&#href_anchor"nav-path" class="navpath">
/usr/share/doc/packages/tesseract-ocr-devel/html/a00014_source.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00014_source.html 2022-03-29 06:12:58.205579719 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00014_source.html 2022-03-29 06:12:58.197579688 +0000
@@ -5,7 +5,7 @@
tesseract: src/ccstruct/werd.cpp Source File
+tesseract: src/cutil/cutil_class.cpp Source File
-
Go to the documentation of this file.
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
19 #href_anchor"code" href="a00095.html">werd.h"
-
-
21 #href_anchor"code" href="a00206.html">linlsq.h"
+
Go to the documentation of this file.
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
19 #href_anchor"code" href="a00023.html">cutil_class.h"
+
+
-
-
-
25 # include "config_auto.h"
-
+
+
+
+
-
28 #define FIRST_COLOUR ScrollView::RED
-
29 #define LAST_COLOUR ScrollView::AQUAMARINE
-
30 #define CHILD_COLOUR ScrollView::BROWN
-
-
-
-
-
43 WERD ::
WERD (C_BLOB_LIST* blob_list, uint8_t blank_count, const
char * text)
-
44 : blanks(blank_count), flags(0), script_id_(0), correct(text) {
-
45 C_BLOB_IT start_it = &cblobs;
-
46 C_BLOB_IT rej_cblob_it = &rej_cblobs;
-
47 C_OUTLINE_IT c_outline_it;
-
48 int16_t inverted_vote = 0;
-
49 int16_t non_inverted_vote = 0;
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
67 start_it.set_to_list(&cblobs);
-
68 if (start_it.empty())
return ;
-
69 for (start_it.mark_cycle_pt(); !start_it.cycled_list(); start_it.forward()) {
-
70 bool reject_blob =
false ;
-
-
-
73 c_outline_it.set_to_list(start_it.data()->out_list());
-
-
75 for (c_outline_it.mark_cycle_pt();
-
76 !c_outline_it.cycled_list() && !reject_blob; c_outline_it.forward()) {
-
77 reject_blob = c_outline_it.data()->flag(
COUT_INVERSE ) != blob_inverted;
-
-
-
80 rej_cblob_it.add_after_then_move(start_it.extract());
-
-
-
-
-
-
-
-
-
89 flags.set_bit(
W_INVERSE , (inverted_vote > non_inverted_vote));
-
-
91 start_it.set_to_list(&cblobs);
-
92 if (start_it.empty())
return ;
-
93 for (start_it.mark_cycle_pt(); !start_it.cycled_list(); start_it.forward()) {
-
94 c_outline_it.set_to_list(start_it.data()->out_list());
-
-
96 rej_cblob_it.add_after_then_move(start_it.extract());
-
-
-
-
-
-
109 : flags(clone->flags),
-
110 script_id_(clone->script_id_),
-
111 correct(clone->correct) {
-
112 C_BLOB_IT start_it = blob_list;
-
113 C_BLOB_IT end_it = blob_list;
-
-
115 while (!end_it.at_last()) end_it.forward();
-
116 (
reinterpret_cast< C_BLOB_LIST*
> (&cblobs))
-
117 ->assign_to_sublist(&start_it, &end_it);
-
-
119 blanks = clone->blanks;
-
-
-
-
-
-
-
126 C_BLOB_LIST temp_blobs;
-
127 C_BLOB_IT temp_it(&temp_blobs);
-
128 temp_it.add_after_then_move(blob);
-
129 WERD * blob_word =
new WERD (&temp_blobs,
this );
-
-
-
-
-
-
-
-
-
-
-
-
154 int bottom = box.
bottom ();
-
-
-
157 C_BLOB_IT it(
const_cast< C_BLOB_LIST*
> (&rej_cblobs));
-
158 for (it.mark_cycle_pt(); !it.cycled_list(); it.forward()) {
-
159 TBOX dot_box = it.data()->bounding_box();
-
160 if ((upper_dots || dot_box.
bottom () <= top) &&
-
161 (lower_dots || dot_box.
top () >= bottom)) {
-
-
-
-
-
-
-
-
-
-
-
172 C_BLOB_IT it(
const_cast< C_BLOB_LIST*
> (&cblobs));
-
173 for (it.mark_cycle_pt(); !it.cycled_list(); it.forward()) {
-
174 box += it.data()->bounding_box();
-
-
-
-
-
-
187 C_BLOB_IT cblob_it(&cblobs);
-
-
189 for (cblob_it.mark_cycle_pt(); !cblob_it.cycled_list(); cblob_it.forward())
-
190 cblob_it.data()->move(vec);
-
-
-
/usr/share/doc/packages/tesseract-ocr-devel/html/a00017.html differs (HTML document, ASCII text)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00017.html 2022-03-29 06:12:58.229579811 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00017.html 2022-03-29 06:12:58.233579825 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/boxword.h File Reference
+
tesseract: src/cutil/callcpp.cpp File Reference
@@ -83,26 +83,360 @@
+
-
#include "genericvector.h "
-
#href_anchor"el" href="a00062_source.html">rect.h"
+
#include "callcpp.h "
+
#href_anchor"el" href="a00854_source.html">scrollview.h"
+
#include "tprintf.h "
Go to the source code of this file.
-class &#href_anchor"memItemRight" valign="bottom">tesseract::BoxWord &#href_anchor"memberdecls">
- namespace tesseract &#href_anchor"nav-path" class="navpath">
+ void&#href_anchor"memItemRight" valign="bottom">cprintf (const char *format,...) ScrollView *&#href_anchor"memItemRight" valign="bottom">c_create_window (const char *name, int16_t xpos, int16_t ypos, int16_t xsize, int16_t ysize, double xmin, double xmax, double ymin, double ymax) void c_line_color_index (void *win, C_COL index) void c_move (void *win, double x, double y) void c_draw (void *win, double x, double y) void c_make_current (void *win) void c_clear_window (void *win) char window_wait (ScrollView *win)&#href_anchor"groupheader">Function Documentation
+◆
+
+
+
+ void c_clear_window
+ (
+ void *
+ win )
+
+
+
+
◆
+
+
+
+ ScrollView * c_create_window (
+ const char *&#href_anchor"paramname">name ,
+
+
+ int16_t
+ xpos ,
+
+ int16_t
+ ypos ,
+
+ int16_t
+ xsize ,
+
+ int16_t
+ ysize ,
+
+ double
+ xmin ,
+
+ double
+ xmax ,
+
+ double
+ ymin ,
+
+ double
+ ymax
+
+ )
+
+
+
+
+
Definition at line 47 of file callcpp.cpp .
+
57 {
+
58 return new ScrollView (name, xpos, ypos, xsize, ysize, xmax - xmin, ymax - ymin,
true );
+
59 }
+
+
+
◆
+
+
+
+ void c_draw
+ (
+ void *
+ win ,
+
+ double
+ x ,
+
+ double
+ y
+
+ )
+
+
+
+
+
Definition at line 80 of file callcpp.cpp .
+
83 {
+
+
85 window->
DrawTo (
static_cast< int > (x),
static_cast< int > (y));
+
86 }
+
void DrawTo(int x, int y)
+
+
+
◆ tesseract: src/ccstruct/boxword.h Source File
+tesseract: src/cutil/callcpp.cpp Source File
-
Go to the documentation of this file.
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
20 #href_anchor"line"> 21 #define TESSERACT_CSTRUCT_BOXWORD_H_
-
-
-
24 #href_anchor"code" href="a00062.html">rect.h"
-
-
-
-
-
+
Go to the documentation of this file.
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
20 #href_anchor"line"> 21 #include "config_auto.h"
+
+
+
+
25 #href_anchor"comment">// for va_end, va_list, va_start
+
+
+
+
29 #href_anchor"code" href="a00398.html">tprintf.h"
-
-
-
-
-
-
-
-
-
-
-
-
-
+
+
+
33 const char *format, ...
+
+
+
+
+
38 va_start(args, format);
+
39 vsprintf(msg, format, args);
+
+
+
+
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
+
+
46 #ifndef GRAPHICS_DISABLED
+
+
+
+
+
+
+
+
+
+
+
+
58 return new ScrollView (name, xpos, ypos, xsize, ysize, xmax - xmin, ymax - ymin,
true );
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
76 window->
SetCursor (
static_cast< int > (x),
static_cast< int > (y));
+
+
-
-
-
-
83 int length ()
const {
return length_; }
-
-
-
+
+
+
+
+
+
85 window->
DrawTo (
static_cast< int > (x),
static_cast< int > (y));
+
-
-
89 void ComputeBoundingBox();
-
-
-
-
-
+
+
+
+
+
+
+
/usr/share/doc/packages/tesseract-ocr-devel/html/a00020.html differs (HTML document, ASCII text)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00020.html 2022-03-29 06:12:58.285580024 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00020.html 2022-03-29 06:12:58.301580086 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/dppoint.cpp File Reference
+
tesseract: src/cutil/callcpp.h File Reference
@@ -83,22 +83,592 @@
+
-
#include "dppoint.h "
-
#href_anchor"el" href="a00647_source.html">errcode.h"
-
#include "tprintf.h "
+
#include "params.h "
+
#href_anchor"el" href="a00482_source.html">unichar.h"
Go to the source code of this file.
-namespace &#href_anchor"memItemRight" valign="bottom">tesseract &#href_anchor"nav-path" class="navpath">
+ enum &#href_anchor"memItemRight" valign="bottom">C_COL { Black
+, White
+, Red
+, Yellow
+, Green
+, Cyan
+, Blue
+, Magenta
+, Aquamarine
+, Dark_SLATE_BLUE
+, Light_BLUE
+, Medium_BLUE
+, Midnight_BLUE
+, Navy_BLUE
+, Sky_BLUE
+, Slate_BLUE
+, Steel_BLUE
+, Coral
+, Brown
+, Sandy_BROWN
+, Gold
+, GoldENROD
+, Dark_GREEN
+, Dark_OLIVE_GREEN
+, Forest_GREEN
+, Lime_GREEN
+, Pale_GREEN
+, Yellow_GREEN
+, Light_GREY
+, Dark_SLATE_GREY
+, Dim_GREY
+, Grey
+, Khaki
+, Maroon
+, Orange
+, Orchid
+, Pink
+, Plum
+, Indian_RED
+, Orange_RED
+, Violet_RED
+, Salmon
+, Tan
+, Turqoise
+, Dark_TURQUOISE
+, Violet
+, Wheat
+, Green_YELLOW
+
+void cprintf (const char *format,...) ScrollView *&#href_anchor"memItemRight" valign="bottom">c_create_window (const char *name, int16_t xpos, int16_t ypos, int16_t xsize, int16_t ysize, double xmin, double xmax, double ymin, double ymax) void c_line_color_index (void *win, C_COL index) void c_move (void *win, double x, double y) void c_draw (void *win, double x, double y) void c_make_current (void *win) void c_clear_window (void *win) char window_wait (ScrollView *win)&#href_anchor"groupheader">Enumeration Type Documentation
+◆
+
+
+Enumerator
+
+
Definition at line 27 of file callcpp.h .
+
27 {
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
Missing in old package: /usr/share/doc/packages/tesseract-ocr-devel/html/a00020.js
Missing in old package: /usr/share/doc/packages/tesseract-ocr-devel/html/a00020.js
/usr/share/doc/packages/tesseract-ocr-devel/html/a00020_source.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00020_source.html 2022-03-29 06:12:58.333580208 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00020_source.html 2022-03-29 06:12:58.341580239 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/dppoint.cpp Source File
+
tesseract: src/cutil/callcpp.h Source File
@@ -82,123 +82,182 @@
Go to the documentation of this file.
-
-
+
+
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
20 #href_anchor"code" href="a00074.html">dppoint.h"
-
-
22 #href_anchor"code" href="a00668.html">tprintf.h"
-
-
-
-
-
-
-
-
-
-
-
33 if (size <= 0 || max_step < min_step || min_step >= size)
-
-
-
-
-
-
-
40 for (
int i = 0; i < size; ++i) {
-
41 for (
int offset = min_step; offset <= max_step; ++offset) {
-
42 DPPoint * prev = offset <= i ? points + i - offset :
nullptr ;
-
43 int64_t new_cost = (points[i].*cost_func)(prev);
-
44 if (points[i].best_prev_ !=
nullptr && offset > min_step * 2 &&
-
45 new_cost > points[i].total_cost_)
-
-
-
48 points[i].total_cost_ += points[i].local_cost_;
-
-
50 tprintf (
"At point %d, local cost=%d, total_cost=%d, steps=%d\n" ,
-
51 i, points[i].local_cost_, points[i].total_cost_,
-
52 points[i].total_steps_);
-
-
-
-
56 int best_cost = points[size - 1].total_cost_;
-
57 int best_end = size - 1;
-
58 for (
int end = best_end - 1; end >= size - min_step; --end) {
-
59 int cost = points[end].total_cost_;
-
60 if (cost < best_cost) {
-
-
-
-
-
65 return points + best_end;
-
-
-
-
-
70 if (prev ==
nullptr || prev ==
this ) {
-
71 UpdateIfBetter(0, 1,
nullptr , 0, 0, 0);
-
-
-
-
75 int delta =
this - prev;
-
76 int32_t n = prev->n_ + 1;
-
77 int32_t sig_x = prev->sig_x_ + delta;
-
78 int64_t sig_xsq = prev->sig_xsq_ + delta * delta;
-
79 int64_t cost = (sig_xsq - sig_x * sig_x / n) / n;
-
80 cost += prev->total_cost_;
-
81 UpdateIfBetter(cost, prev->total_steps_ + 1, prev, n, sig_x, sig_xsq);
-
-
-
-
-
86 void DPPoint::UpdateIfBetter(int64_t cost, int32_t steps,
const DPPoint * prev,
-
87 int32_t n, int32_t sig_x, int64_t sig_xsq) {
-
88 if (cost < total_cost_) {
-
-
-
-
-
-
-
-
-
-
-
-
DLLSYM void tprintf(const char *format,...)
-
-
-
-
-
-
static DPPoint * Solve(int min_step, int max_step, bool debug, CostFunc cost_func, int size, DPPoint *points)
-
int64_t CostWithVariance(const DPPoint *prev)
-
int64_t(DPPoint::*)(const DPPoint *) CostFunc
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
19 #href_anchor"line"> 20 #define CALLCPP_H
+
+
+
23 #href_anchor"code" href="a00482.html">unichar.h"
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
/usr/share/doc/packages/tesseract-ocr-devel/html/a00023.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00023.html 2022-03-29 06:12:58.365580330 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00023.html 2022-03-29 06:12:58.365580330 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/polyblk.cpp File Reference
+
tesseract: src/cutil/cutil_class.h File Reference
@@ -83,100 +83,26 @@
+
-
#include "polyblk.h "
-
#href_anchor"el" href="a00665_source.html">elst.h"
-
#include <cctype>
-
#include <cinttypes>
-
#include <cmath>
-
#include <cstdio>
-
#include <memory>
+
#include "ccutil.h "
+
#href_anchor"el" href="a00395_source.html">strngs.h"
Go to the source code of this file.
-#href_anchor"memItemRight" valign="bottom">INTERSECTING INT16_MAX
-Reflect the coords of the polygon in the y-axis. (Flip the sign of x.)
-
int lessthan (const void *first, const void *second)
-
-
-
◆
-
-
-
-
-
- #define INTERSECTING INT16_MAX
-
-
-
-
-
-
-
◆
-
-
-
-
-
- int lessthan
- (
- const void *
- first ,
-
- const void *
- second
-
- )
-
-
-
-
-
Definition at line 379 of file polyblk.cpp .
-
379 {
-
-
-
382
-
383 if (p1->
x () < p2->
x ())
-
384 return (-1);
-
385 else if (p1->
x () > p2->
x ())
-
386 return (1);
-
387 else
-
388 return (0);
-
389 }
-
int16_t x() const
access function
-
-
-
-
-
-
-
-
+
+
class &#href_anchor"memItemRight" valign="bottom">tesseract::CUtil
+
&#href_anchor"memberdecls">
+
+
namespace tesseract
+
&#href_anchor"nav-path" class="navpath">
/usr/share/doc/packages/tesseract-ocr-devel/html/a00023_source.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00023_source.html 2022-03-29 06:12:58.397580453 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00023_source.html 2022-03-29 06:12:58.385580407 +0000
@@ -5,7 +5,7 @@
tesseract: src/ccstruct/polyblk.cpp Source File
+tesseract: src/cutil/cutil_class.h Source File
-
Go to the documentation of this file.
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
18 #href_anchor"code" href="a00056.html">polyblk.h"
-
-
20 #href_anchor"line"> 21 #include <cinttypes>
-
-
-
-
-
-
-
28 #include "config_auto.h"
-
-
-
31 #define INTERSECTING INT16_MAX
-
-
33 int lessthan (
const void *first,
const void *second);
-
-
-
36 ICOORDELT_IT v = &vertices;
-
-
-
-
-
-
-
-
-
-
-
-
48 ICOORDELT_IT v = &vertices;
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
69 ICOORDELT_IT pts = &vertices;
-
-
71 botleft = *pts.data ();
-
-
-
-
75 if (pos.
x () < botleft.
x ())
-
-
77 botleft =
ICOORD (pos.
x (), botleft.
y ());
-
78 if (pos.
y () < botleft.
y ())
-
79 botleft =
ICOORD (botleft.
x (), pos.
y ());
-
80 if (pos.
x () > topright.
x ())
-
81 topright =
ICOORD (pos.
x (), topright.
y ());
-
82 if (pos.
y () > topright.
y ())
-
83 topright =
ICOORD (topright.
x (), pos.
y ());
-
-
-
86 while (!pts.at_first ());
-
87 ibl =
ICOORD (botleft.
x (), botleft.
y ());
-
88 itr =
ICOORD (topright.
x (), topright.
y ());
-
89 box =
TBOX (ibl, itr);
-
-
-
-
-
-
-
-
-
-
106 ICOORDELT_IT it = &vertices;
-
-
-
-
-
-
112 vvec = *it.data_relative (1) - pt;
-
-
114 if (vec.
y () <= 0 && vec.
y () + vvec.
y () > 0) {
-
-
-
-
-
-
-
121 else if (vec.
y () > 0 && vec.
y () + vvec.
y () <= 0) {
-
-
-
-
-
-
-
128 else if (vec.
y () == 0 && vec.
x () == 0)
-
-
-
-
132 while (!it.at_first ());
-
-
-
-
-
-
-
140 ICOORDELT_IT it = &vertices;
-
-
-
-
-
-
-
-
-
149 vertex = *it.data ();
-
-
-
-
-
-
-
-
157 while (!it.at_first ());
-
-
-
-
-
162 it.set_to_list (other->
points ());
-
-
164 vertex = *it.data ();
-
-
-
-
-
-
-
-
172 while (!it.at_first ());
-
-
-
-
-
-
-
-
187 ICOORDELT_IT pts = &vertices;
-
-
-
-
-
-
-
194 pt->
set_x (
static_cast< int16_t
> (floor(pos.
x () + 0.5)));
-
195 pt->
set_y (
static_cast< int16_t
> (floor(pos.
y () + 0.5)));
-
-
-
198 while (!pts.at_first ());
-
-
-
-
-
/usr/share/doc/packages/tesseract-ocr-devel/html/a00026.html differs (HTML document, ASCII text)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00026.html 2022-03-29 06:12:58.421580544 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00026.html 2022-03-29 06:12:58.425580560 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/detlinefit.h File Reference
+
tesseract: src/cutil/oldlist.cpp File Reference
@@ -83,27 +83,491 @@
+
-
#include "genericvector.h "
-
#href_anchor"el" href="a00656_source.html">kdpair.h"
-
#include "points.h "
+
#include "oldlist.h "
+
#href_anchor"el" href="a00446_source.html">errcode.h"
Go to the source code of this file.
+
+
+
◆
+
+
+
+
+
+ #define copy_first
+ (
+
+ l1,
+
+
+
+ l2
+
+
+ )
+ (l2=push (l2, first_node (l1)))
+
+
+
+
+
+
+
◆
+
+
+
+
+
+ int count
+ (
+ LIST var_list )
+
+
+
+
+
Definition at line 95 of file oldlist.cpp .
+
95 {
+
96 int temp = 0;
+
97
+
+
99 return (temp);
+
100 }
+
+
+
+
+
+
◆
+
+
+
+
+
+ LIST delete_d (
+ LIST list ,
+
+ void *
+ key ,
+
+ int_compare is_equal
+
+ )
+
+
+
+
+
Definition at line 110 of file oldlist.cpp .
+
110 {
+
+
+
113
+
+
115
+
+
+
+
119 last_one = list;
+
+
121 result = last_one;
+
+
123 } else {
+
+
125 last_one = list;
+
+
+
128 }
+
129 } else {
+
+
131 }
+
132 }
+
133 return (result);
+
134 }
+
#define set_rest(l, cell)
+
+
+
+
+
+
+
+
+
+
+
◆
+
+
Missing in old package: /usr/share/doc/packages/tesseract-ocr-devel/html/a00026.js
Missing in old package: /usr/share/doc/packages/tesseract-ocr-devel/html/a00026.js
/usr/share/doc/packages/tesseract-ocr-devel/html/a00026_source.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00026_source.html 2022-03-29 06:12:58.453580667 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00026_source.html 2022-03-29 06:12:58.453580667 +0000
@@ -5,7 +5,7 @@
tesseract: src/ccstruct/detlinefit.h Source File
+tesseract: src/cutil/oldlist.cpp Source File
-
Go to the documentation of this file.
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
20 #href_anchor"line"> 21 #define TESSERACT_CCSTRUCT_DETLINEFIT_H_
-
-
-
24 #href_anchor"code" href="a00656.html">kdpair.h"
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
71 void Add (
const ICOORD & pt,
int halfwidth);
-
-
-
-
-
76 return Fit (0, 0, pt1, pt2);
-
-
-
-
-
-
-
-
-
-
-
-
-
-
90 double min_dist,
double max_dist,
-
91 bool debug,
ICOORD * line_pt);
-
-
-
-
-
-
-
-
-
100 double Fit (
float * m,
float * c);
+
Go to the documentation of this file.
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
59 #href_anchor"line"> 60 #include <cstring>
+
+
+
+
+
+
+
+
+
+
+
71 #define copy_first(l1,l2) \
+
72 (l2=push(l2, first_node(l1)))
+
+
+
+
+
+
+
+
+
+
+
+
+
+
86 static int is_same(
void *item1,
void *item2) {
/usr/share/doc/packages/tesseract-ocr-devel/html/a00029.html differs (HTML document, ASCII text)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00029.html 2022-03-29 06:12:58.477580759 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00029.html 2022-03-29 06:12:58.477580759 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/boxword.cpp File Reference
+
tesseract: src/dict/trie.cpp File Reference
@@ -85,15 +85,16 @@
-
+
-
#include "blobs.h "
-
#href_anchor"el" href="a00017_source.html">boxword.h"
-
#include "host.h "
-
#href_anchor"el" href="a00032_source.html">normalis.h"
-
#include "ocrblock.h "
-
#href_anchor"el" href="a00227_source.html">pageres.h"
+
#include "trie.h "
+
#href_anchor"el" href="a00020_source.html">callcpp.h"
+
#include "dawg.h "
+
#href_anchor"el" href="a00062_source.html">dict.h"
+
#include "genericvector.h "
+
#href_anchor"el" href="a00512_source.html">helpers.h"
+
#include "kdpair.h "
Go to the source code of this file.
/usr/share/doc/packages/tesseract-ocr-devel/html/a00029.js differs (ASCII text)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00029.js 2020-03-26 00:00:00.000000000 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00029.js 2020-03-26 00:00:00.000000000 +0000
@@ -1,4 +1,7 @@
var a00029 =
[
- [ "kBoxClipTolerance", "a00029.html#a03af8c63d879bb74722a52f7bee8c5de", null ]
+ [ "kDoNotReverse", "a00029.html#a0db2198747060995d61a01dcfac97eb7", null ],
+ [ "kForceReverse", "a00029.html#a597d22620e40ecc0ec38dfa4f65a9d85", null ],
+ [ "kReverseIfHasRTL", "a00029.html#a402a13eb2d1593daed0190a118f23d88", null ],
+ [ "RTLReversePolicyNames", "a00029.html#af5c5a40d574f6ab56726b8b2f963f65c", null ]
];
\ No newline at end of file
/usr/share/doc/packages/tesseract-ocr-devel/html/a00029_source.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00029_source.html 2022-03-29 06:12:58.525580942 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00029_source.html 2022-03-29 06:12:58.545581018 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/boxword.cpp Source File
+
tesseract: src/dict/trie.cpp Source File
@@ -82,283 +82,870 @@
-
Go to the documentation of this file.
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
19 #href_anchor"code" href="a00068.html">blobs.h"
-
-
21 #href_anchor"code" href="a00653.html">host.h"
-
-
23 #href_anchor"code" href="a00179.html">ocrblock.h"
-
+
Go to the documentation of this file.
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
24 #href_anchor"code" href="a00059.html">trie.h"
-
-
-
-
-
-
+
+
27 #href_anchor"code" href="a00035.html">dawg.h"
+
+
29 #href_anchor"code" href="a00392.html">genericvector.h"
+
+
31 #href_anchor"code" href="a00530.html">kdpair.h"
-
-
-
-
-
-
-
-
-
-
-
+
+
+
+
+
+
+
+
+
+
+
-
-
-
47 length_ = src.length_;
-
-
-
50 for (
int i = 0; i < length_; ++i)
-
-
-
-
-
-
-
-
-
59 boxword->length_ = tessword->
NumBlobs ();
-
-
61 boxword->boxes_.reserve(boxword->length_);
-
-
63 for (
int b = 0; b < boxword->length_; ++b) {
-
-
-
-
67 outline = outline->
next ) {
-
68 EDGEPT * edgept = outline->loop;
-
-
-
-
-
-
-
-
-
77 TBOX pt_box(pos, pos);
-
-
-
80 edgept = edgept->
next ;
-
81 }
while (edgept != outline->loop);
-
-
83 boxword->boxes_.push_back(blob_box);
-
-
85 boxword->ComputeBoundingBox();
-
-
-
-
-
-
-
-
93 for (
int i = 0; i < length_; ++i) {
-
-
-
-
-
-
-
-
101 for (b_it.mark_cycle_pt(); !b_it.cycled_list(); b_it.forward()) {
-
102 TBOX blob_box = b_it.data()->bounding_box();
-
103 if (block !=
nullptr )
-
-
-
106 original_box += blob_box;
-
-
-
-
-
-
112 if (NearlyEqual<int>(original_box.
right (), box.
right (),
-
-
-
-
-
117 if (NearlyEqual<int>(original_box.
bottom (), box.
bottom (),
-
-
-
-
-
122 if (block !=
nullptr )
-
-
-
-
126 ComputeBoundingBox();
-
-
-
-
-
-
-
-
134 if (end <= start + 1)
-
-
136 for (
int i = start + 1; i < end; ++i) {
-
137 boxes_[start] += boxes_[i];
-
-
139 int shrinkage = end - 1 - start;
-
140 length_ -= shrinkage;
-
141 for (
int i = start + 1; i < length_; ++i)
/usr/share/doc/packages/tesseract-ocr-devel/html/a00032.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00032.html 2022-03-29 06:12:58.565581095 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00032.html 2022-03-29 06:12:58.565581095 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/normalis.h File Reference
+
tesseract: src/dict/dict.cpp File Reference
@@ -83,96 +83,23 @@
+
-
+
#include <cstdio>
+
#include "dict.h "
+
#href_anchor"el" href="a00422_source.html">unicodes.h"
+
#include "tprintf.h "
+
Go to the source code of this file.
-
-
-
◆
-
-
-
-
-
- const int kBlnBaselineOffset = 64
-
-
-
-
-
-
◆
-
-
-
-
-
- const int kBlnCellHeight = 256
-
-
-
-
-
-
◆
-
-
-
-
-
- const int kBlnXHeight = 128
-
-
-
-
-
-
-
-
+
namespace &#href_anchor"memItemRight" valign="bottom">tesseract
+
&#href_anchor"nav-path" class="navpath">
/usr/share/doc/packages/tesseract-ocr-devel/html/a00032_source.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00032_source.html 2022-03-29 06:12:58.605581248 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00032_source.html 2022-03-29 06:12:58.621581310 +0000
@@ -5,7 +5,7 @@
tesseract: src/ccstruct/normalis.h Source File
+tesseract: src/dict/dict.cpp Source File
-
Go to the documentation of this file.
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
20 #href_anchor"line"> 21 #define NORMALIS_H
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
177 float x_origin,
float y_origin,
-
-
179 float final_xshift,
float final_yshift);
-
-
-
/usr/share/doc/packages/tesseract-ocr-devel/html/a00035.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00035.html 2022-03-29 06:12:58.649581416 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00035.html 2022-03-29 06:12:58.653581432 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/statistc.cpp File Reference
+
tesseract: src/dict/dawg.h File Reference
@@ -83,284 +83,309 @@
+
-
#include "statistc.h "
-
#href_anchor"el" href="a00647_source.html">errcode.h"
-
#include "helpers.h "
-
#href_anchor"el" href="a00794_source.html">scrollview.h"
-
#include "tprintf.h "
+
#include <cinttypes>
+
#include <memory>
+
#include "elst.h "
+
#href_anchor"el" href="a00371_source.html">params.h"
+
#include "ratngs.h "
+
#href_anchor"el" href="a00515_source.html">tesscallback.h"
Go to the source code of this file.
-int32_t&#href_anchor"memItemRight" valign="bottom">choose_nth_item (int32_t index, float *array, int32_t count ) int32_t choose_nth_item (int32_t index, void *array, int32_t count , size_t size, int(*compar)(const void *, const void *)) void swap_entries (void *array, size_t size, int32_t index1, int32_t index2) struct &#href_anchor"memItemRight" valign="bottom">tesseract::NodeChild &#href_anchor"memitem:"> class tesseract::Dawg &#href_anchor"memitem:"> struct tesseract::DawgPosition &#href_anchor"memitem:"> class tesseract::DawgPositionVector &#href_anchor"memitem:"> class tesseract::SquishedDawg &#href_anchor"memberdecls">
+ namespace tesseract &#href_anchor"memberdecls">
+ #define FORWARD_EDGE (int32_t) 0 #define BACKWARD_EDGE (int32_t) 1 #define MAX_NODE_EDGES_DISPLAY (int64_t) 100 #define MARKER_FLAG (int64_t) 1 #define DIRECTION_FLAG (int64_t) 2 #define WERD_END_FLAG (int64_t) 4 #define LETTER_START_BIT 0 #define NUM_FLAG_BITS 3 #define REFFORMAT "%" PRId64
-
-
-
◆ [1/2]
-
-
-
-
-
- int32_t choose_nth_item
- (
- int32_t
- index ,
-
- float *
- array ,
-
- int32_t
- count
-
- )
-
-
-
-
-
Definition at line 630 of file statistc.cpp .
-
630 {
-
631 int32_t next_sample;
-
632 int32_t next_lesser;
-
633 int32_t prev_greater;
-
634 int32_t equal_count;
-
635 float pivot;
-
-
637
-
-
639 return 0;
-
-
641 if (array[0] < array[1]) {
-
642 return index >= 1 ? 1 : 0;
-
643 }
-
644 else {
-
645 return index >= 1 ? 0 : 1;
-
646 }
-
647 }
-
648 else {
-
649 if (index < 0)
-
650 index = 0;
-
651 else if (index >=
count )
-
-
653 equal_count =
static_cast< int32_t
> (rand() %
count );
-
654 pivot = array[equal_count];
-
655
-
656 array[equal_count] = array[0];
-
657 next_lesser = 0;
-
658 prev_greater =
count ;
-
659 equal_count = 1;
-
660 for (next_sample = 1; next_sample < prev_greater;) {
-
661 sample = array[next_sample];
-
-
663
-
664 array[next_lesser++] =
sample ;
-
665 next_sample++;
-
666 }
-
667 else if (
sample > pivot) {
-
668 prev_greater--;
-
669
-
670 array[next_sample] = array[prev_greater];
-
671 array[prev_greater] =
sample ;
-
672 }
-
673 else {
-
674 equal_count++;
-
675 next_sample++;
-
676 }
-
677 }
-
678 for (next_sample = next_lesser; next_sample < prev_greater;)
-
679 array[next_sample++] = pivot;
-
680 if (index < next_lesser)
-
-
682 else if (index < prev_greater)
-
683 return next_lesser;
-
684 else
-
-
686 array + prev_greater,
/usr/share/doc/packages/tesseract-ocr-devel/html/a00035.js differs (ASCII text)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00035.js 2020-03-26 00:00:00.000000000 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00035.js 2020-03-26 00:00:00.000000000 +0000
@@ -1,6 +1,32 @@
var a00035 =
[
- [ "choose_nth_item", "a00035.html#adf07d3ffc714e6fd2731a59c34a00bc4", null ],
- [ "choose_nth_item", "a00035.html#a8d581ad9304be945cf9e5193593fc607", null ],
- [ "swap_entries", "a00035.html#abf6a19728e788edfabb5ec4f3c61bfce", null ]
+ [ "tesseract::NodeChild", "a03561.html", "a03561" ],
+ [ "tesseract::Dawg", "a03565.html", "a03565" ],
+ [ "tesseract::DawgPosition", "a03569.html", "a03569" ],
+ [ "tesseract::DawgPositionVector", "a03573.html", "a03573" ],
+ [ "tesseract::SquishedDawg", "a03577.html", "a03577" ],
+ [ "BACKWARD_EDGE", "a00035.html#a1b595d17c69b114ee1042e96e51a19bd", null ],
+ [ "DIRECTION_FLAG", "a00035.html#afc8f59f76bf176bfc7ee72f7f96bcc7a", null ],
+ [ "FORWARD_EDGE", "a00035.html#a1e19d6a48aa61d422cce885b812bb733", null ],
+ [ "LETTER_START_BIT", "a00035.html#abd290e0ea4e54e330815019aae6148aa", null ],
+ [ "MARKER_FLAG", "a00035.html#a7e0df79bf5779569d0c9f325216e7613", null ],
+ [ "MAX_NODE_EDGES_DISPLAY", "a00035.html#a2d04619cb05b140c086f2ec13778d235", null ],
+ [ "NUM_FLAG_BITS", "a00035.html#af4e00364d8957bd37d5f641879ecc37a", null ],
+ [ "REFFORMAT", "a00035.html#a85a5bf216156dfa399c5edd26f3694db", null ],
+ [ "WERD_END_FLAG", "a00035.html#a0fe1bde7916f6e5869319634b341449d", null ],
+ [ "EDGE_ARRAY", "a00035.html#a55706b9e674564b826bbd93f03812f5d", null ],
+ [ "EDGE_RECORD", "a00035.html#ac49a2017de57ddd4abd4ff07e91e2e4b", null ],
+ [ "EDGE_REF", "a00035.html#aff972a2dc9ce96f0507f40ac1fc57ab7", null ],
+ [ "NODE_MAP", "a00035.html#a45c76558fd622726402917010e263036", null ],
+ [ "NODE_REF", "a00035.html#a2dae42e8033c5e1a0abd1bb90446a105", null ],
+ [ "NodeChildVector", "a00035.html#a48a16e430d43b82af15134dfa4d42840", null ],
+ [ "SuccessorList", "a00035.html#afbcb88ad726dd6494791b98926e6d1da", null ],
+ [ "SuccessorListsVector", "a00035.html#a4ba77fa8c5c5ee1c2067ccd8d9568c37", null ],
+ [ "DawgType", "a00035.html#a07c41f251f1782c60df4e626c35d2151", [
+ [ "DAWG_TYPE_PUNCTUATION", "a00035.html#a07c41f251f1782c60df4e626c35d2151aa8e8b65bf32d2884c0bab5fb8223c6eb", null ],
+ [ "DAWG_TYPE_WORD", "a00035.html#a07c41f251f1782c60df4e626c35d2151a9c67e397ff6d57a705cd4cd4524a2b34", null ],
+ [ "DAWG_TYPE_NUMBER", "a00035.html#a07c41f251f1782c60df4e626c35d2151a16f2414a05db0ec4b7e1b266052a966f", null ],
+ [ "DAWG_TYPE_PATTERN", "a00035.html#a07c41f251f1782c60df4e626c35d2151a4e994eccb48eec56bb31bccd8bbb3cef", null ],
+ [ "DAWG_TYPE_COUNT", "a00035.html#a07c41f251f1782c60df4e626c35d2151ac823345423c28fc2793bfdb0db9f9942", null ]
+ ] ]
];
\ No newline at end of file
/usr/share/doc/packages/tesseract-ocr-devel/html/a00035_source.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00035_source.html 2022-03-29 06:12:58.705581631 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00035_source.html 2022-03-29 06:12:58.713581662 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/statistc.cpp Source File
+
tesseract: src/dict/dawg.h Source File
@@ -82,830 +82,595 @@
-
Go to the documentation of this file.
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
20 #href_anchor"line"> 21 #include "config_auto.h"
-
-
-
-
25 #href_anchor"line"> 26 #include <cmath>
-
-
-
29 #href_anchor"code" href="a00635.html">helpers.h"
-
-
31 #href_anchor"code" href="a00668.html">tprintf.h"
-
-
-
-
-
-
-
-
-
40 STATS::STATS (int32_t min_bucket_value, int32_t max_bucket_value_plus_1) {
-
41 if (max_bucket_value_plus_1 <= min_bucket_value) {
-
-
43 max_bucket_value_plus_1 = 1;
-
-
45 rangemin_ = min_bucket_value;
-
46 rangemax_ = max_bucket_value_plus_1;
-
47 buckets_ =
new int32_t[rangemax_ - rangemin_];
-
-
-
-
-
-
-
-
-
-
57 if (max_bucket_value_plus_1 <= min_bucket_value) {
-
-
-
60 if (rangemax_ - rangemin_ != max_bucket_value_plus_1 - min_bucket_value) {
-
-
62 buckets_ =
new int32_t[max_bucket_value_plus_1 - min_bucket_value];
-
-
64 rangemin_ = min_bucket_value;
-
65 rangemax_ = max_bucket_value_plus_1;
-
-
-
-
-
-
-
-
-
-
-
-
77 if (buckets_ !=
nullptr )
-
78 memset(buckets_, 0, (rangemax_ - rangemin_) *
sizeof (buckets_[0]));
-
+
Go to the documentation of this file.
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
22 #href_anchor"line"> 23 #define DICT_DAWG_H_
+
+
+
+
+
+
+
+
+
32 #href_anchor"code" href="a00371.html">params.h"
+
+
34 #href_anchor"code" href="a00515.html">tesscallback.h"
+
+
+
+
38 #define NO_EDGE (int64_t) 0xffffffffffffffffi64
+
+
+
41 #define NO_EDGE (int64_t) 0xffffffffffffffffll
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
-
-
-
-
-
-
-
-
-
-
-
-
-
-
94 if (buckets_ ==
nullptr ) {
-
-
-
97 value =
ClipToRange (value, rangemin_, rangemax_ - 1);
-
98 buckets_[value - rangemin_] +=
count ;
-
99 total_count_ +=
count ;
-
-
-
-
-
-
/usr/share/doc/packages/tesseract-ocr-devel/html/a00038.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00038.html 2022-03-29 06:12:58.741581768 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00038.html 2022-03-29 06:12:58.741581768 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/blamer.cpp File Reference
+
tesseract: src/dict/stopper.h File Reference
@@ -83,283 +83,57 @@
+
-
#include "blamer.h "
-
#href_anchor"el" href="a00068_source.html">blobs.h"
-
#include "errcode.h "
-
#href_anchor"el" href="a01202_source.html">lm_pain_points.h"
-
#include "matrix.h "
-
#href_anchor"el" href="a00032_source.html">normalis.h"
-
#include "pageres.h "
-
#href_anchor"el" href="a00770_source.html">tesscallback.h"
-
#include "unicharset.h "
+
#include "genericvector.h "
+
#href_anchor"el" href="a00371_source.html">params.h"
+
#include "ratngs.h "
+
#href_anchor"el" href="a00482_source.html">unichar.h"
Go to the source code of this file.
-
-
-
◆
-
-
-
-
-
- const char kBlameAdaption[] = "adapt"
-
-
-
-
-
-
◆
-
-
-
-
-
- const char kBlameChopper[] = "chop"
-
-
-
-
-
-
◆
-
-
-
-
-
- const char kBlameClassifier[] = "cl"
-
-
-
-
-
-
◆
-
-
-
-
-
- const char kBlameClassLMTradeoff[] = "cl/LM"
-
-
-
-
-
-
◆
-
-
-
-
-
- const char kBlameClassOldLMTradeoff[] = "cl/old_LM"
-
-
-
-
-
-
◆
-
-
-
-
-
- const char kBlameCorrect[] = "corr"
-
-
-
-
-
-
◆
-
-
-
-
-
- const char kBlameNoTruth[] = "no_tr"
-
-
-
-
-
-
◆
-
-
-
-
-
- const char kBlameNoTruthSplit[] = "no_tr_spl"
-
-
-
-
-
-
◆
-
-
/usr/share/doc/packages/tesseract-ocr-devel/html/a00038.js differs (ASCII text)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00038.js 2020-03-26 00:00:00.000000000 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00038.js 2020-03-26 00:00:00.000000000 +0000
@@ -1,16 +1,6 @@
var a00038 =
[
- [ "kBlameAdaption", "a00038.html#afbe9ed9b199de04329403be78d43a20e", null ],
- [ "kBlameChopper", "a00038.html#a4db1a8ed37b0eff0d2e038ff2a107c2c", null ],
- [ "kBlameClassifier", "a00038.html#a0691e45d7e7b5e6e0671f5167e4a0e71", null ],
- [ "kBlameClassLMTradeoff", "a00038.html#a5c0e7d2e8cba382e6de905c1612dad3b", null ],
- [ "kBlameClassOldLMTradeoff", "a00038.html#aa05d635a577a5fafa7fd3b2a3ef0c17d", null ],
- [ "kBlameCorrect", "a00038.html#a939a828ea97a292eed52468bfc886f5f", null ],
- [ "kBlameNoTruth", "a00038.html#acbc066860637c1eb6ba9777ec6914020", null ],
- [ "kBlameNoTruthSplit", "a00038.html#af5f1678db33a085c6a72c606584c8f6f", null ],
- [ "kBlamePageLayout", "a00038.html#ad7fc57611e72df3bbe8a760f7d9a8bf1", null ],
- [ "kBlameSegsearchHeur", "a00038.html#a89e8d8eda8453e09c7b0c95d60a57f7c", null ],
- [ "kBlameSegsearchPP", "a00038.html#a5dc22321ca86643deb20076a2198c8cf", null ],
- [ "kBlameUnknown", "a00038.html#a5c9260acc89f21815ea4f8c8fc12e473", null ],
- [ "kIncorrectResultReasonNames", "a00038.html#a85fc3ba2e9db3adeb717fbdba6f9481f", null ]
+ [ "DANGERR_INFO", "a03605.html", "a03605" ],
+ [ "BLOB_WIDTH", "a00038.html#af5f37aa612be5b5d3563816d12defe94", null ],
+ [ "DANGERR", "a00038.html#a83aebc66a5c063bfae114275effe34b2", null ]
];
\ No newline at end of file
/usr/share/doc/packages/tesseract-ocr-devel/html/a00038_source.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00038_source.html 2022-03-29 06:12:58.789581952 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00038_source.html 2022-03-29 06:12:58.777581906 +0000
@@ -5,7 +5,7 @@
tesseract: src/ccstruct/blamer.cpp Source File
+tesseract: src/dict/stopper.h Source File
-
Go to the documentation of this file.
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
21 #href_anchor"code" href="a00197.html">blamer.h"
-
-
-
-
25 #href_anchor"code" href="a00647.html">errcode.h"
-
26 #if !defined(DISABLED_LEGACY_ENGINE)
-
-
28 #href_anchor"line"> 29 #include "matrix.h "
-
30 #href_anchor"code" href="a00032.html">normalis.h"
-
-
32 #href_anchor"code" href="a00770.html">tesscallback.h"
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
+
Go to the documentation of this file.
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
18 #href_anchor"line"> 19 #define STOPPER_H
+
+
+
+
+
+
+
26 #href_anchor"code" href="a00371.html">params.h"
+
+
28 #href_anchor"code" href="a00482.html">unichar.h"
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
75 const char * truth_str,
const TBOX & word_box) {
-
-
77 truth_has_char_boxes_ =
false ;
-
-
-
-
81 unicharset.
encode_string (truth_str,
false , &encoding, &lengths,
nullptr );
-
-
83 for (
int i = 0; i < encoding.
size (); total_length += lengths[i++]) {
-
84 STRING uch(truth_str + total_length);
-
-
-
-
-
-
-
-
-
-
-
95 const char * char_str,
const TBOX & char_box) {
-
96 STRING symbol_str(char_str);
-
-
98 if (
id != INVALID_UNICHAR_ID) {
-
-
100 if (normed_uch.
length () > 0) symbol_str = normed_uch;
-
-
102 int length = truth_word_.
length ();
-
-
-
-
106 truth_has_char_boxes_ =
true ;
-
107 else if (truth_word_.
BlobBox (length - 1) == char_box)
-
108 truth_has_char_boxes_ =
false ;
-
-
-
-
-
-
-
115 truth_has_char_boxes_ =
false ;
-
-
-
-
-
120 if (word_choice ==
nullptr )
return false ;
-
-
-
123 for (
int i = 0; i < word_choice->
length (); ++i) {
-
-
-
-
-
128 return truth_str == normed_choice_str;
-
-
-
-
-
-
134 (*debug) +=
"Truth " ;
-
135 for (
int i = 0; i < this->truth_text_.
length (); ++i) {
-
136 (*debug) += this->truth_text_[i];
-
/usr/share/doc/packages/tesseract-ocr-devel/html/a00041.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00041.html 2022-03-29 06:12:58.817582059 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00041.html 2022-03-29 06:12:58.813582043 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/blobbox.cpp File Reference
+
tesseract: src/dict/dawg.cpp File Reference
@@ -83,638 +83,26 @@
+
-
#include "blobbox.h "
-
#href_anchor"el" href="a00068_source.html">blobs.h"
-
#include "coutln.h "
-
#href_anchor"el" href="a00635_source.html">helpers.h"
-
#include "host.h "
-
#href_anchor"el" href="a00158_source.html">points.h"
+
#include "dawg.h "
+
#href_anchor"el" href="a00062_source.html">dict.h"
+
#include "helpers.h "
+
#href_anchor"el" href="a00395_source.html">strngs.h"
+
#include "tesscallback.h "
+
#href_anchor"el" href="a00398_source.html">tprintf.h"
+
#include <memory>
Go to the source code of this file.
-void find_cblob_limits (C_BLOB *blob, float leftx, float rightx, FCOORD rotation, float &ymin, float &ymax)&#href_anchor"memitem:a57197f6f9deea830c41b3af8c9b8abb5"> void find_cblob_vlimits (C_BLOB *blob, float leftx, float rightx, float &ymin, float &ymax)&#href_anchor"memitem:a3a16fa420feca9ee9f2211e31de0f0dd"> void find_cblob_hlimits (C_BLOB *blob, float bottomy, float topy, float &xmin, float &xmax)&#href_anchor"memitem:abf66d3434cfbea9a4b988aea4416e4a3"> C_BLOB *&#href_anchor"memItemRight" valign="bottom">crotate_cblob (C_BLOB *blob, FCOORD rotation)&#href_anchor"memitem:a343c30f9e935058965b844710a2e3180"> TBOX &#href_anchor"memItemRight" valign="bottom">box_next (BLOBNBOX_IT *it) TBOX &#href_anchor"memItemRight" valign="bottom">box_next_pre_chopped (BLOBNBOX_IT *it) void vertical_cblob_projection (C_BLOB *blob, STATS *stats)&#href_anchor"memitem:a48333b2ce7f8f0d732fc47a82e57911c"> void vertical_coutline_projection (C_OUTLINE *outline, STATS *stats)&#href_anchor"memitem:a331e6744e04cb0a2fb23c1465c82196c"> void plot_blob_list (ScrollView *win, BLOBNBOX_LIST *list, ScrollView::Color body_colour, ScrollView::Color child_colour)
-
-
-
◆
-
-
-
-
-
- #define PROJECTION_MARGIN 10
-
-
-
-
-
-
-
◆
-
-
-
-
-
- TBOX box_next (
- BLOBNBOX_IT *&#href_anchor"paramname">it )
-
-
-
-
-
Definition at line 636 of file blobbox.cpp .
-
638 {
-
-
-
641
-
642 blob = it->data ();
-
-
644 do {
-
645 it->forward ();
-
646 blob = it->data ();
-
647 if (blob->
cblob () ==
nullptr )
-
648
-
-
650 }
-
651
-
-
653 return result;
-
654 }
-
-
const TBOX & bounding_box() const
-
-
bool joined_to_prev() const
-
-
-
-
-
-
◆
-
-
-
-
-
- TBOX box_next_pre_chopped (
- BLOBNBOX_IT *&#href_anchor"paramname">it )
-
-
-
-
-
Definition at line 665 of file blobbox.cpp .
-
667 {
-
-
-
670
-
671 blob = it->data ();
-
-
673 do {
-
674 it->forward ();
-
675 blob = it->data ();
-
676 }
-
677
-
-
679 return result;
-
680 }
-
-
-
-
-
◆
-
-
-
-
-
- C_BLOB * crotate_cblob (
- C_BLOB *&#href_anchor"paramname">blob ,
-
- FCOORD &#href_anchor"paramname">rotation
-
- )
-
-
-
-
-
Definition at line 611 of file blobbox.cpp .
-
614 {
-
615 C_OUTLINE_LIST out_list;
-
616
-
617 C_OUTLINE_IT in_it = blob->
out_list ();
-
618
-
619 C_OUTLINE_IT out_it = &out_list;
-
620
-
621 for (in_it.mark_cycle_pt (); !in_it.cycled_list (); in_it.forward ()) {
-
622 out_it.add_after_then_move (
new C_OUTLINE (in_it.data (), rotation));
-
623 }
-
624 return new C_BLOB (&out_list);
/usr/share/doc/packages/tesseract-ocr-devel/html/a00041_source.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00041_source.html 2022-03-29 06:12:58.877582288 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00041_source.html 2022-03-29 06:12:58.861582227 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/blobbox.cpp Source File
+
tesseract: src/dict/dawg.cpp Source File
@@ -82,1319 +82,504 @@
-
Go to the documentation of this file.
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
20 #href_anchor"line"> 21 #include "config_auto.h"
-
+
Go to the documentation of this file.
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
-
-
25 #href_anchor"comment">// for max, min
-
-
27 #include "allheaders.h"
-
-
29 #href_anchor"code" href="a00215.html">coutln.h"
-
-
-
32 #href_anchor"code" href="a00653.html">host.h"
-
-
-
-
-
37 #define PROJECTION_MARGIN 10
+
24 #href_anchor"code" href="a00035.html">dawg.h"
+
+
+
27 #href_anchor"code" href="a00512.html">helpers.h"
+
+
29 #href_anchor"code" href="a00515.html">tesscallback.h"
+
+
+
+
+
+
+
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
56 cblob_ptr->
rotate (rotation);
-
-
58 compute_bounding_box();
-
-
-
-
-
63 int left = -box.
right ();
-
-
-
-
-
-
-
-
-
-
-
-
75 ICOORD bottom_pt(top_pt.
x (), base_char_bottom_);
-
-
77 base_char_top_ = top_pt.
y ();
-
78 bottom_pt.
rotate (rotation);
-
79 base_char_bottom_ = bottom_pt.
y ();
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
97 nextblob->joined =
true ;
-
-
-
-
-
-
-
104 if (other->cblob_ptr !=
nullptr ) {
-
105 C_OUTLINE_IT ol_it(cblob_ptr->
out_list ());
-
106 ol_it.add_list_after(other->cblob_ptr->
out_list ());
-
-
+
+
+
+
+
+
+
45 bool requires_complete)
const {
+
46 if (word.
length () == 0)
return !requires_complete;
+
+
48 int end_index = word.
length () - 1;
+
49 for (
int i = 0; i < end_index; i++) {
+
+
51 if (edge == NO_EDGE) {
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
71 bool enable_wildcard)
const {
+
72 if (filename ==
nullptr )
return 0;
+
+
+
+
+
/usr/share/doc/packages/tesseract-ocr-devel/html/a00044.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00044.html 2022-03-29 06:12:58.901582380 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00044.html 2022-03-29 06:12:58.901582380 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/linlsq.cpp File Reference
+
tesseract: src/dict/hyphen.cpp File Reference
@@ -83,52 +83,20 @@
+
-
#include <cstdio>
-
#include <cmath>
-
#include "errcode.h "
-
#href_anchor"el" href="a00206_source.html">linlsq.h"
+
Go to the source code of this file.
-constexpr ERRCODE &#href_anchor"memItemRight" valign="bottom">EMPTY_LLSQ ("Can't delete from an empty LLSQ ") &#href_anchor"groupheader">Function Documentation
-◆
-
-
-
-
-
-
- constexpr ERRCODE EMPTY_LLSQ
- (
- "Can't delete from an empty LLSQ "&#href_anchor"paramname"> )
-
-
-
-
-constexpr
-
-
-
-
-
-
+
+
namespace &#href_anchor"memItemRight" valign="bottom">tesseract
+
&#href_anchor"nav-path" class="navpath">
/usr/share/doc/packages/tesseract-ocr-devel/html/a00044_source.html differs (HTML document, UTF-8 Unicode text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00044_source.html 2022-03-29 06:12:58.925582472 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00044_source.html 2022-03-29 06:12:58.925582472 +0000
@@ -5,7 +5,7 @@
tesseract: src/ccstruct/linlsq.cpp Source File
+tesseract: src/dict/hyphen.cpp Source File
-
Go to the documentation of this file.
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
19 #href_anchor"line"> 20 #include <cmath>
-
-
22 #href_anchor"code" href="a00206.html">linlsq.h"
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
58 total_weight += weight;
-
-
-
61 sigxx += x * x * weight;
-
62 sigxy += x * y * weight;
-
63 sigyy += y * y * weight;
-
-
-
-
67 total_weight += other.total_weight;
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
83 if (total_weight <= 0.0)
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
104 return covar / x_var;
-
-
-
-
-
-
-
-
-
-
-
-
-
117 if (total_weight > 0.0)
-
118 return (sigy -
m * sigx) / total_weight;
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
133 if (total_weight > 0) {
-
134 error = sigyy +
m * (
m * sigxx + 2 * (
c * sigx - sigxy)) +
c *
-
135 (total_weight *
c - 2 * sigy);
-
-
137 error = std::sqrt(error / total_weight);
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
159 if (var_product > 0.0)
-
160 r = covar / std::sqrt(var_product);
-
-
-
-
-
-
-
167 if (total_weight > 0.0) {
-
168 return FCOORD (sigx / total_weight, sigy / total_weight);
-
-
170 return FCOORD (0.0f, 0.0f);
-
-
-
-
-
-
-
-
-
-
-
-
/usr/share/doc/packages/tesseract-ocr-devel/html/a00047.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00047.html 2022-03-29 06:12:58.969582640 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00047.html 2022-03-29 06:12:58.953582579 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/polyaprx.h File Reference
+
tesseract: src/dict/permdawg.cpp File Reference
@@ -83,702 +83,26 @@
+
-
+
#include "dawg.h "
+
#href_anchor"el" href="a00038_source.html">stopper.h"
+
#include "tprintf.h "
+
#href_anchor"el" href="a00371_source.html">params.h"
+
#include <algorithm>
+
#include <cctype>
+
#include "dict.h "
+
Go to the source code of this file.
-TESSLINE *&#href_anchor"memItemRight" valign="bottom">ApproximateOutline (bool allow_detailed_fx, C_OUTLINE *c_outline)&#href_anchor"memitem:aff4b13bfd414b463da57ae26acc0c6f3"> EDGEPT *&#href_anchor"memItemRight" valign="bottom">edgesteps_to_edgepts (C_OUTLINE *c_outline, EDGEPT edgepts[])&#href_anchor"memitem:a90ebde0147c2a36a0b1a216645a4eed8"> void fix2 (EDGEPT *start, int area)&#href_anchor"memitem:ab3ec2f619a85e4d8174d3c54084705f7"> EDGEPT *&#href_anchor"memItemRight" valign="bottom">poly2 (EDGEPT *startpt, int area)&#href_anchor"memitem:aa0f03bd75099af8feca58259b9f3f48d"> void cutline (EDGEPT *first, EDGEPT *last , int area)
-
-
-
◆
-
-
-
-
-
- TESSLINE * ApproximateOutline (
- bool&#href_anchor"paramname">allow_detailed_fx ,
-
-
- C_OUTLINE *&#href_anchor"paramname">c_outline
-
- )
-
-
-
-
-
Definition at line 61 of file polyaprx.cpp .
-
61 {
-
-
63 int32_t area;
-
-
65 EDGEPT * edgepts = stack_edgepts;
-
66
-
67
-
-
-
70
-
-
-
73 if (!poly_wide_objects_better && loop_box.
width () > area)
-
74 area = loop_box.
width ();
-
75 area *= area;
-
-
-
-
-
-
81 EDGEPT * prev_result =
nullptr ;
-
82 do {
-
-
-
85 new_pt->prev = prev_result;
-
86 if (prev_result == nullptr ) {
-
87 result = new_pt;
-
88 } else {
-
89 prev_result->
next = new_pt;
-
90 new_pt->
prev = prev_result;
-
91 }
-
92 if (allow_detailed_fx) {
-
-
-
-
96 }
-
97 prev_result = new_pt;
-
98 edgept = edgept->
next ;
-
99 }
-
100 while (edgept != startpt);
-
101 prev_result->
next = result;
-
102 result->
prev = prev_result;
-
103 if (edgepts != stack_edgepts)
-
104 delete [] edgepts;
-
-
106 }
-
void fix2(EDGEPT *start, int area)
-
EDGEPT * poly2(EDGEPT *startpt, int area)
-
-
EDGEPT * edgesteps_to_edgepts(C_OUTLINE *c_outline, EDGEPT edgepts[])
-
-
-
-
-
-
-
-
static TESSLINE * BuildFromOutlineList(EDGEPT *outline)
-
const TBOX & bounding_box() const
-
int32_t pathlength() const
-
-
-
-
-
-
-
-
◆
-
-
-
-
-
- void cutline
- (
- EDGEPT *&#href_anchor"paramname">first ,
-
- EDGEPT *&#href_anchor"paramname">last ,
-
- int
- area
-
- )
-
-
-
-
-
Definition at line 498 of file polyaprx.cpp .
-
502 {
-
-
-
505 int vlen;
-
-
-
508 int maxperp;
-
509 int perp;
-
510 int ptcount;
-
511 int squaresum;
-
512
-
513 edge = first;
-
-
515 return ;
-
516
-
517
-
-
-
520 if (vecsum.
x == 0 && vecsum.
y == 0) {
-
521
-
-
-
524 }
-
525
-
526 vlen = vecsum.
x > 0 ? vecsum.
x : -vecsum.
x ;
-
-
-
529 else if (-vecsum.
y > vlen)
-
-
531
-
-
-
534 maxperp = 0;
-
535 squaresum = ptcount = 0;
-
-
537 maxpoint = edge;
/usr/share/doc/packages/tesseract-ocr-devel/html/a00047_source.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00047_source.html 2022-03-29 06:12:58.993582732 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00047_source.html 2022-03-29 06:12:59.009582793 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/polyaprx.h Source File
+
tesseract: src/dict/permdawg.cpp Source File
@@ -82,69 +82,421 @@
-
Go to the documentation of this file.
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
20 #href_anchor"line"> 21 #define POLYAPRX_H
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
TESSLINE * ApproximateOutline(bool allow_detailed_fx, C_OUTLINE *c_outline)
-
void fix2(EDGEPT *start, int area)
-
void cutline(EDGEPT *first, EDGEPT *last, int area)
-
EDGEPT * poly2(EDGEPT *startpt, int area)
-
EDGEPT * edgesteps_to_edgepts(C_OUTLINE *c_outline, EDGEPT edgepts[])
-
-
-
-
-
+
Go to the documentation of this file.
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
24 #href_anchor"code" href="a00035.html">dawg.h"
+
+
26 #href_anchor"code" href="a00398.html">tprintf.h"
+
+
+
29 #href_anchor"line"> 30 #include <cctype>
+
+
+
+
+
+
+
+
+
+
+
47 bool word_ending,
WERD_CHOICE *word,
float certainties[],
float *limit,
+
48 WERD_CHOICE *best_choice,
int *attempts_left,
void *void_more_args) {
+
49 auto *more_args =
static_cast< DawgArgs *
> (void_more_args);
+
50 word_ending = (char_choice_index == char_choices.
size ()-1);
+
51 int word_index = word->
length () - 1;
+
52 if (best_choice->
rating () < *limit)
return ;
+
+
+
+
+
+
58 bool checked_unigrams =
false ;
+
+
+
61 tprintf (
"checking unigrams in an ngram %s\n" ,
+
+
+
+
+
+
+
+
+
+
71 bool unigrams_ok =
true ;
+
+
+
+
75 DawgArgs unigram_dawg_args(&unigram_active_dawgs,
+
76 &unigram_updated_dawgs,
+
+
+
79 for (
int i = 0; unigrams_ok && i < encoding.
size (); ++i) {
+
+
+
+
+
+
+
+
87 word_ending && i == encoding.
size () - 1);
+
+
+
+
+
92 unigrams_ok ?
"OK" :
"not OK" );
+
+
+
+
+
+
+
99 checked_unigrams =
true ;
+
100 more_args->permuter = unigram_dawg_args.
permuter ;
+
101 *(more_args->updated_dawgs) = *(unigram_dawg_args.
updated_dawgs );
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
116 if (output_ambig_words_file_ ==
nullptr ) {
+
117 output_ambig_words_file_ =
+
+
119 if (output_ambig_words_file_ ==
nullptr ) {
+
120 tprintf (
"Failed to open output_ambig_words_file %s\n" ,
+
+
+
+
+
+
+
127 fprintf(output_ambig_words_file_,
"%s" , word_str.
string ());
+
+
+
+
+
132 fprintf(output_ambig_words_file_,
"%s" , word_str.
string ());
+
+
/usr/share/doc/packages/tesseract-ocr-devel/html/a00050.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00050.html 2022-03-29 06:12:59.025582854 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00050.html 2022-03-29 06:12:59.029582869 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/mod128.h File Reference
+
tesseract: src/dict/context.cpp File Reference
@@ -83,83 +83,31 @@
+
-
#include "points.h "
+
#include "dict.h "
+
#href_anchor"el" href="a00431_source.html">unicharset.h"
Go to the source code of this file.
-
-
-
◆
-
-
-
-
-
- #define DIRBITS 7
-
-
-
-
-
-
◆
-
-
-
-
-
- #define DIRSCALE 1000
-
-
-
-
-
-
◆
-
-
-
-
-
- #define MODULUS 128 /*range of directions */
-
-
-
-
/usr/share/doc/packages/tesseract-ocr-devel/html/a00050.js differs (ASCII text)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00050.js 2020-03-26 00:00:00.000000000 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00050.js 2020-03-26 00:00:00.000000000 +0000
@@ -1,7 +1,4 @@
var a00050 =
[
- [ "DIR128", "a02769.html", "a02769" ],
- [ "DIRBITS", "a00050.html#a657b731308eeb83f8a113f73968e1d4a", null ],
- [ "DIRSCALE", "a00050.html#a73610b3264640201a811ce96e3ffaad9", null ],
- [ "MODULUS", "a00050.html#a9bc1fe94bf8c1b9aaef3dade106b6304", null ]
+ [ "case_state_table", "a00050.html#ae599808f0375f76eb5e9340ec448d7f2", null ]
];
\ No newline at end of file
/usr/share/doc/packages/tesseract-ocr-devel/html/a00050_source.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00050_source.html 2022-03-29 06:12:59.061582993 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00050_source.html 2022-03-29 06:12:59.065583007 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/mod128.h Source File
+
tesseract: src/dict/context.cpp Source File
@@ -82,108 +82,108 @@
-
Go to the documentation of this file.
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
20 #href_anchor"line"> 21 #define MOD128_H
-
-
-
-
-
-
+
Go to the documentation of this file.
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
21 #href_anchor"code" href="a00062.html">dict.h"
+
+
+
+
+
26 static const int kMinAbsoluteGarbageWordLength = 10;
+
27 static const float kMinAbsoluteGarbageAlphanumFrac = 0.5f;
-
-
-
-
-
-
-
-
-
-
-
39 dir =
static_cast< int8_t
> (value);
-
-
-
-
-
-
-
-
-
48 dir =
static_cast< int8_t
> (value);
-
-
-
-
52 const DIR128 & minus)
const
-
-
-
55 int16_t result = dir - minus.dir;
-
-
-
-
-
-
61 return static_cast< int8_t
> (result);
-
-
-
-
-
-
-
68 result = dir + add.dir;
-
-
-
-
-
73 *
this = dir + add.dir;
-
-
-
-
-
-
-
-
-
-
-
-
-
ICOORD & operator+=(ICOORD &op1, const ICOORD &op2)
-
ICOORD operator-(const ICOORD &src)
-
ICOORD operator+(const ICOORD &op1, const ICOORD &op2)
-
-
-
-
-
-
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
50 for (x = 0; x < word.
length (); ++x) {
+
+
+
+
+
+
+
+
+
+
60 if (state == -1)
return false ;
+
+
+
+
+
+
+
67 if (word.
length () < kMinAbsoluteGarbageWordLength)
return false ;
+
+
69 for (
int x = 0; x < word.
length (); ++x) {
+
+
+
+
73 return (
static_cast< float > (num_alphanum) /
+
74 static_cast< float > (word.
length ()) < kMinAbsoluteGarbageAlphanumFrac);
+
+
+
+
+
+
+
+
const int case_state_table[6][4]
+
+
UNICHAR_ID unichar_id(int index) const
+
const UNICHARSET * unicharset() const
+
+
+
bool get_isupper(UNICHAR_ID unichar_id) const
+
bool get_isalpha(UNICHAR_ID unichar_id) const
/usr/share/doc/packages/tesseract-ocr-devel/html/a00053.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00053.html 2022-03-29 06:12:59.093583114 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00053.html 2022-03-29 06:12:59.089583099 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/split.cpp File Reference
+
tesseract: src/dict/stopper.cpp File Reference
@@ -83,231 +83,33 @@
+
-
#include "split.h "
-
#href_anchor"el" href="a00215_source.html">coutln.h"
-
#include "tprintf.h "
-
#href_anchor"a00053_source.html">Go to the source code of this file.
-
-
-◆
-
-
-
- EDGEPT * make_edgept (
- int&#href_anchor"paramname">x ,
-
-
- int
- y ,
-
- EDGEPT *&#href_anchor"paramname">next ,
-
- EDGEPT *&#href_anchor"paramname">prev
-
- )
-
-
-
-
-
Definition at line 139 of file split.cpp .
-
139 {
-
-
141
-
-
143 this_edgept->
pos .
x = x;
-
144 this_edgept->
pos .
y = y;
-
145
-
-
147 if (prev_ol !=
nullptr && prev->
next == next) {
-
148
-
-
-
151 double cut_fraction = target_vec.length() / segment_vec.length();
-
152
-
-
-
-
-
157 ICOORD step_vec = step_end - step_start;
-
158 double target_length = step_vec.
length () * cut_fraction;
-
159
-
-
-
162 double best_dist = target_length;
-
163 for (
int s = prev->
start_step ; s < end_step; ++s) {
-
164 total_step += prev_ol->
step (s % step_length);
-
165 double dist = fabs(target_length - total_step.length());
-
166 if (dist < best_dist) {
-
167 best_dist = dist;
-
168 best_step = s + 1;
-
169 }
-
170 }
-
171
-
-
173 this_edgept->
step_count = end_step - best_step;
-
174 this_edgept->
start_step = best_step % step_length;
-
-
176 } else {
-
177
-
-
-
-
181 }
-
182
-
183 this_edgept->
next = next;
-
184 this_edgept->
prev = prev;
-
185 prev->
next = this_edgept;
-
186 next->
prev = this_edgept;
-
187
-
-
-
-
-
192 return this_edgept;
-
193 }
-
-
-
-
-
-
-
-
-
-
-
-
ICOORD step(int index) const
-
ICOORD position_at_index(int index) const
-
int32_t pathlength() const
-
-
float length() const
find length
-
-
-
-
◆
-
-
-
- void remove_edgept
- (
- EDGEPT *&#href_anchor"paramname">point )
-
-
-
-
-
Definition at line 200 of file split.cpp .
-
200 {
-
-
-
203
-
-
-
206 }
-
-
-
-
-
211 delete point;
-
212 }
-
-
-
◆
-
-
-
- const double kBadPriority = 999.0
-
-
-
-
-
Definition at line 39 of file split.cpp .
-
/usr/share/doc/packages/tesseract-ocr-devel/html/a00053_source.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00053_source.html 2022-03-29 06:12:59.121583221 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00053_source.html 2022-03-29 06:12:59.133583267 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/split.cpp Source File
+
tesseract: src/dict/stopper.cpp Source File
@@ -82,401 +82,632 @@
-
Go to the documentation of this file.
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
23 #href_anchor"line"> 24 #include "config_auto.h"
-
-
-
-
28 #href_anchor"code" href="a00215.html">coutln.h"
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
55 edgept = edgept->
next ;
-
-
-
-
-
60 edgept = edgept->
next ;
-
-
-
-
-
-
-
-
-
69 edgept = edgept->
next ;
-
-
-
-
-
74 edgept = edgept->
next ;
-
-
-
-
-
-
-
-
82 int centered_maxwidth,
double center_knob,
-
83 double width_change_knob)
const {
-
-
-
86 int min_left = std::min(box1.
left (), box2.
left ());
-
87 int max_right = std::max(box1.
right (), box2.
right ());
-
88 if (xmin < min_left && xmax > max_right)
return kBadPriority ;
-
-
-
-
92 int width1 = box1.
width ();
-
93 int width2 = box2.
width ();
-
94 int min_width = std::min(width1, width2);
-
95 int overlap = -box1.
x_gap (box2);
-
96 if (overlap == min_width) {
-
-
-
99 if (2 * overlap > min_width) overlap += 2 * overlap - min_width;
-
100 if (overlap > 0) grade += overlap_knob * overlap;
-
-
-
103 if (width1 <= centered_maxwidth || width2 <= centered_maxwidth) {
-
104 grade += std::min(
static_cast< double > (
kCenterGradeCap ), center_knob * abs(width1 - width2));
-
-
-
107 float width_change_grade = 20 - (max_right - min_left - std::max(width1, width2));
-
108 if (width_change_grade > 0.0f)
-
109 grade += width_change_grade * width_change_knob;
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
143 this_edgept->
pos .
x = x;
-
144 this_edgept->
pos .
y = y;
-
-
-
147 if (prev_ol !=
nullptr && prev->
next == next) {
-
-
-
-
151 double cut_fraction = target_vec.
length () / segment_vec.
length ();
-
-
-
-
-
-
157 ICOORD step_vec = step_end - step_start;
-
158 double target_length = step_vec.
length () * cut_fraction;
-
-
-
-
162 double best_dist = target_length;
-
163 for (
int s = prev->
start_step ; s < end_step; ++s) {
-
164 total_step += prev_ol->
step (s % step_length);
-
165 double dist = fabs(target_length - total_step.
length ());
-
166 if (dist < best_dist) {
-
-
-
-
-
-
-
173 this_edgept->
step_count = end_step - best_step;
-
174 this_edgept->
start_step = best_step % step_length;
-
-
-
-
-
-
-
-
-
183 this_edgept->
next = next;
/usr/share/doc/packages/tesseract-ocr-devel/html/a00056.html differs (HTML document, ASCII text)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00056.html 2022-03-29 06:12:59.157583359 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00056.html 2022-03-29 06:12:59.157583359 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/polyblk.h File Reference
+
tesseract: src/dict/dawg_cache.h File Reference
@@ -83,25 +83,28 @@
+
-
#include "publictypes.h "
-
#href_anchor"el" href="a00665_source.html">elst.h"
-
#include "points.h "
-
#href_anchor"el" href="a00062_source.html">rect.h"
-
#include "scrollview.h "
+
#include "dawg.h "
+
#href_anchor"el" href="a00437_source.html">object_cache.h"
+
#include "strngs.h "
+
#href_anchor"el" href="a00449_source.html">tessdatamanager.h"
Go to the source code of this file.
class &#href_anchor"memItemRight" valign="bottom">POLY_BLOCK &#href_anchor"memitem:"> class PB_LINE_IT class &#href_anchor"memItemRight" valign="bottom">tesseract::DawgCache &#href_anchor"memberdecls">
+ namespace tesseract &#href_anchor"nav-path" class="navpath">
/usr/share/doc/packages/tesseract-ocr-devel/html/a00056_source.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00056_source.html 2022-03-29 06:12:59.181583451 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00056_source.html 2022-03-29 06:12:59.185583466 +0000
@@ -5,7 +5,7 @@
tesseract: src/ccstruct/polyblk.h Source File
+tesseract: src/dict/dawg_cache.h Source File
-
Go to the documentation of this file.
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
18 #href_anchor"line"> 19 #define POLYBLK_H
-
-
-
22 #href_anchor"code" href="a00665.html">elst.h"
-
-
24 #href_anchor"code" href="a00062.html">rect.h"
-
-
-
-
-
-
-
-
-
+
Go to the documentation of this file.
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
20 #href_anchor"line"> 21 #define TESSERACT_DICT_DAWG_CACHE_H_
+
+
+
24 #href_anchor"code" href="a00437.html">object_cache.h"
+
+
26 #href_anchor"code" href="a00449.html">tessdatamanager.h"
+
+
+
+
+
+
+
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
+
+
+
+
+
39 return dawgs_.Free(dawg);
+
+
+
+
+
44 dawgs_.DeleteUnusedObjects();
+
+
+
+
+
+
+
-
-
-
55 void rotate(
FCOORD rotation);
-
-
57 void reflect_in_y_axis();
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
76 int16_t winding_number(
const ICOORD &test_pt);
-
-
78 #href_anchor"line"> 79
-
-
-
-
-
-
85 ICOORDELT_LIST vertices;
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
105 ICOORDELT_LIST *get_line(int16_t y);
-
-
-
-
-
110 #href_anchor"ttc" id="aa00062_html">
-
-
-
bool PTIsTextType(PolyBlockType type)
-
-
-
-
-
-
-
-
-
-
ICOORDELT_LIST * points()
-
-
-
PolyBlockType isA() const
-
-
-
void set_to_block(POLY_BLOCK *blkptr)
-
PB_LINE_IT(POLY_BLOCK *blkptr)
-
-
-
+
53 #href_anchor"comment">// TESSERACT_DICT_DAWG_CACHE_H_
+
+
/usr/share/doc/packages/tesseract-ocr-devel/html/a00059.html differs (HTML document, ASCII text)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00059.html 2022-03-29 06:12:59.205583542 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00059.html 2022-03-29 06:12:59.209583558 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/quadratc.h File Reference
+
tesseract: src/dict/trie.h File Reference
@@ -83,20 +83,106 @@
+
-
#include "points.h "
+
#include "dawg.h "
+
#href_anchor"el" href="a00392_source.html">genericvector.h"
Go to the source code of this file.
class &#href_anchor"memItemRight" valign="bottom">QUAD_COEFFS &#href_anchor"nav-path" class="navpath">
+ struct &#href_anchor"memItemRight" valign="bottom">TRIE_NODE_RECORD &#href_anchor"memitem:"> class tesseract::Trie &#href_anchor"memberdecls">
+ namespace tesseract &#href_anchor"memberdecls">
+ using EDGE_INDEX = int64_t using NODE_MARKER = bool * using EDGE_VECTOR = GenericVector < EDGE_RECORD > using TRIE_NODES = GenericVector < TRIE_NODE_RECORD * >&#href_anchor"groupheader">Typedef Documentation
+◆
+
+
+
Definition at line 38 of file trie.h .
+
+
+
◆
+
+
+
Definition at line 40 of file trie.h .
+
+
+
◆
+
+
+
Definition at line 39 of file trie.h .
+
+
+
◆
+
+
+
Definition at line 46 of file trie.h .
+
+
+
tesseract: src/ccstruct/quadratc.h Source File
+tesseract: src/dict/trie.h Source File
-
Go to the documentation of this file.
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
20 #href_anchor"line"> 21 #define QUADRATC_H
+
Go to the documentation of this file.
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
20 #href_anchor"line"> 21 #define TRIE_H
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
40 return static_cast< float > ((
a * x +
b ) * x +
c );
-
-
-
-
-
-
-
-
-
-
-
-
-
53 c =
static_cast< float > (
c -
b * p +
a * p * p + q);
-
54 b =
static_cast< float > (
b - 2 *
a * p);
-
-
-
-
-
-
-
-
-
-
-
int16_t y() const
access_function
-
int16_t x() const
access function
-
-
-
-
-
-
QUAD_COEFFS(double xsq, float x, float constant)
-
-
+
+
24 #href_anchor"code" href="a00392.html">genericvector.h"
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
84 int unicharset_size,
int debug_level)
+
+
86 init (unicharset_size);
+
+
+
+
+
+
+
+
+
+
+
+
99 bool word_end)
const override {
+
+
+
+
103 &edge_ptr, &edge_index))
return NO_EDGE;
+
+
+
+
+
112 bool word_end)
const override {
+
+
114 nodes_ [
static_cast< int > (node)]->forward_edges;
+
115 for (
int i = 0; i < forward_edges.
size (); ++i) {
+
+
+
+
+
+
/usr/share/doc/packages/tesseract-ocr-devel/html/a00062.html differs (HTML document, ASCII text)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00062.html 2022-03-29 06:12:59.277583818 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00062.html 2022-03-29 06:12:59.277583818 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/rect.h File Reference
+
tesseract: src/dict/dict.h File Reference
@@ -83,27 +83,111 @@
+
-
#include <algorithm>
-
#include <cmath>
-
#include <cstdint>
-
#include <cstdio>
-
#include "platform.h "
-
#href_anchor"el" href="a00158_source.html">points.h"
-
#include "scrollview.h "
-
#href_anchor"el" href="a00668_source.html">tprintf.h"
+
#include "ambigs.h "
+
#href_anchor"el" href="a00035_source.html">dawg.h"
+
#include "dawg_cache.h "
+
#href_anchor"el" href="a00809_source.html">ratngs.h"
+
#include "stopper.h "
+
#href_anchor"el" href="a00059_source.html">trie.h"
+
#include "unicharset.h "
+
#href_anchor"el" href="a00653_source.html">params_training_featdef.h"
Go to the source code of this file.
+
+
+
◆
+
+
+
+
+
+ #define CHARS_PER_LINE 500
+
+
+
+
+
Definition at line 38 of file dict.h .
+
+
+
+
+
◆
+
+
+
+
+
+ #define MAX_WERD_LENGTH (int64_t) 128
+
+
+
+
+
Definition at line 39 of file dict.h .
+
+
+
+
+
◆
+
+
+
+
+
+ #define NO_RATING -1
+
+
+
+
+
Definition at line 40 of file dict.h .
+
+
+
+
+
+
+
Missing in old package: /usr/share/doc/packages/tesseract-ocr-devel/html/a00062.js
Missing in old package: /usr/share/doc/packages/tesseract-ocr-devel/html/a00062.js
/usr/share/doc/packages/tesseract-ocr-devel/html/a00062_source.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00062_source.html 2022-03-29 06:12:59.317583971 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00062_source.html 2022-03-29 06:12:59.329584016 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/rect.h Source File
+
tesseract: src/dict/dict.h Source File
@@ -82,571 +82,731 @@
-
Go to the documentation of this file.
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
20 #href_anchor"line"> 21 #define RECT_H
-
-
-
-
-
-
-
28 #href_anchor"code" href="a00158.html">points.h"
-
-
30 #href_anchor"code" href="a00668.html">tprintf.h"
-
-
-
-
-
-
-
37 bot_left (INT16_MAX, INT16_MAX), top_right (-INT16_MAX, -INT16_MAX) {
-
-
-
-
-
-
-
-
45 int16_t left, int16_t bottom, int16_t right, int16_t top);
-
-
-
-
-
-
51 return ((left () >= right ()) || (top () <= bottom ()));
-
-
-
-
55 return bot_left == other.bot_left && top_right == other.top_right;
-
-
-
-
59 return top_right.y ();
-
-
-
-
-
-
-
-
-
-
-
+
Go to the documentation of this file.
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
19 #href_anchor"line"> 20 #define TESSERACT_DICT_DICT_H_
+
+
22 #ifndef DISABLED_LEGACY_ENGINE
+
+
24 #href_anchor"line"> 25 #include "dawg.h "
+
26 #href_anchor"code" href="a00056.html">dawg_cache.h"
+
+
28 #href_anchor"code" href="a00038.html">stopper.h"
+
+
30 #href_anchor"code" href="a00431.html">unicharset.h"
+
31 #ifndef DISABLED_LEGACY_ENGINE
+
+
33 #href_anchor"comment">// ndef DISABLED_LEGACY_ENGINE
+
+
+
+
+
38 #define CHARS_PER_LINE 500
+
39 #define MAX_WERD_LENGTH (int64_t) 128
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
58 static const int kRatingPad = 4;
+
59 static const char kDictWildcard[] =
"\u2606" ;
+
60 static const int kDictMaxWildcards = 2;
+
+
+
63 static const char kHyphenSymbol[] =
"-" ;
+
64 static const char kSlashSymbol[] =
"/" ;
+
65 static const char kQuestionSymbol[] =
"?" ;
+
66 static const char kApostropheSymbol[] =
"&#href_anchor"line"> 67 static const float kSimCertaintyScale = -10.0;
+
68 static const float kSimCertaintyOffset = -10.0;
+
69 static const float kSimilarityFloor = 100.0;
+
70 static const int kDocDictMaxRepChars = 4;
-
-
-
-
-
-
-
-
-
80 return top_right.x ();
-
-
-
-
-
-
86 return (bot_left.x() + top_right.x()) / 2;
-
-
-
89 return (bot_left.y() + top_right.y()) / 2;
-
-
-
-
-
-
-
-
97 return ICOORD (top_right.x (), bot_left.y ());
-
-
-
-
101 return ICOORD (bot_left.x (), top_right.y ());
-
-
-
-
-
-
-
-
-
110 return top_right.y () - bot_left.y ();
-
-
-
-
-
-
-
117 return top_right.x () - bot_left.x ();
/usr/share/doc/packages/tesseract-ocr-devel/html/a00065.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00065.html 2022-03-29 06:12:59.349584093 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00065.html 2022-03-29 06:12:59.353584108 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/seam.cpp File Reference
+
tesseract: src/dict/dawg_cache.cpp File Reference
@@ -82,81 +82,30 @@
-
#include "seam.h "
-
#href_anchor"el" href="a00068_source.html">blobs.h"
-
#include "tprintf.h "
+
#include "dawg_cache.h "
+
#href_anchor"el" href="a00035_source.html">dawg.h"
+
#include "object_cache.h "
+
#href_anchor"el" href="a00395_source.html">strngs.h"
+
#include "tessdatamanager.h "
Go to the source code of this file.
-Initialize a list of seams that match the original number of blobs present in the starting segmentation. Each of the seams created by this routine have location information only.
-
void&#href_anchor"memItemRight" valign="bottom">start_seam_list (TWERD *word, GenericVector < SEAM * > *seam_array) &#href_anchor"groupheader">Function Documentation
-◆
-
-
-
- void start_seam_list
- (
- TWERD *&#href_anchor"paramname">word ,
-
- GenericVector < SEAM * > *&#href_anchor"paramname">seam_array
-
- )
-
-
-
-
-
Definition at line 263 of file seam.cpp .
-
263 {
-
-
-
266
-
267 for (
int b = 1; b < word->
NumBlobs (); ++b) {
-
268 TBOX bbox = word->
blobs [b - 1]->bounding_box();
-
269 TBOX nbox = word->
blobs [b]->bounding_box();
-
270 location.
x = (bbox.
right () + nbox.
left ()) / 2;
-
-
-
273 }
-
274 }
-
-
-
-
-
-
-
GenericVector< TBLOB * > blobs
-
-
-
-
-
-
-
-
-
+
+
struct &#href_anchor"memItemRight" valign="bottom">tesseract::DawgLoader
+
&#href_anchor"memberdecls">
+
+
namespace tesseract
+
&#href_anchor"nav-path" class="navpath">
/usr/share/doc/packages/tesseract-ocr-devel/html/a00065_source.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00065_source.html 2022-03-29 06:12:59.377584200 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00065_source.html 2022-03-29 06:12:59.377584200 +0000
@@ -5,7 +5,7 @@
tesseract: src/ccstruct/seam.cpp Source File
+tesseract: src/dict/dawg_cache.cpp Source File
-
Go to the documentation of this file.
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
22 #href_anchor"code" href="a00083.html">seam.h"
-
-
24 #href_anchor"code" href="a00668.html">tprintf.h"
-
-
-
-
-
-
-
-
32 TBOX box(location_.
x , location_.
y , location_.
x , location_.
y );
-
33 for (
int s = 0; s < num_splits_; ++s) {
-
-
-
-
+
Go to the documentation of this file.
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
20 #href_anchor"code" href="a00056.html">dawg_cache.h"
+
+
+
23 #href_anchor"code" href="a00437.html">object_cache.h"
+
+
25 #href_anchor"code" href="a00449.html">tessdatamanager.h"
+
+
+
+
+
+
+
+
+
+
+
+
-
-
-
41 float max_total_priority)
const {
-
42 int dist = location_.
x - other.location_.
x ;
-
43 if (-max_x_dist < dist && dist < max_x_dist &&
-
44 num_splits_ + other.num_splits_ <= kMaxNumSplits &&
-
45 priority_ + other.priority_ < max_total_priority &&
-
-
-
-
-
-
-
-
-
-
55 priority_ += other.priority_;
-
56 location_ += other.location_;
-
-
-
59 for (uint8_t s = 0; s < other.num_splits_ && num_splits_ < kMaxNumSplits; ++s)
-
60 splits_[num_splits_++] = other.splits_[s];
-
-
-
-
-
-
-
-
-
69 return num_splits_ == 0 || splits_[0].
IsHealthy (blob, min_points, min_area);
-
-
-
-
-
-
-
-
-
78 int insert_index,
bool modify) {
-
79 for (
int s = 0; s < insert_index; ++s) {
-
-
-
82 if (!
FindBlobWidth (blobs, insert_index, modify))
return false ;
-
83 for (
int s = insert_index; s < seams.
size (); ++s) {
-
84 if (!seams[s]->
FindBlobWidth (blobs, s + 1, modify))
return false ;
-
-
-
-
-
-
-
-
-
-
-
-
-
-
98 for (
int s = 0; s < num_splits_; ++s) {
-
99 const SPLIT & split = splits_[s];
-
-
-
102 for (
int b = index + 1; !found_split && b < blobs.
size (); ++b) {
-
-
104 if (found_split && b - index > widthp_ && modify) widthp_ = b - index;
-
-
-
107 for (
int b = index - 1; !found_split && b >= 0; --b) {
-
-
109 if (found_split && index - b > widthn_ && modify) widthn_ = index - b;
-
-
111 if (found_split) ++num_found;
-
-
113 return num_found == num_splits_;
-
-
-
-
-
-
119 for (
int s = 0; s < num_splits_; ++s) {
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
141 while (outline->
next ) outline = outline->
next ;
-
-
-
-
/usr/share/doc/packages/tesseract-ocr-devel/html/a00068.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00068.html 2022-03-29 06:12:59.405584307 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00068.html 2022-03-29 06:12:59.409584323 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/blobs.h File Reference
+
tesseract: src/dict/matchdefs.h File Reference
@@ -87,235 +87,427 @@
Macros |
Typedefs |
Functions
-
+
-
#include <cstdint>
-
#include <cstring>
-
#include "clst.h "
-
#href_anchor"el" href="a00758_source.html">genericvector.h"
-
#include "normalis.h "
-
#href_anchor"el" href="a00158_source.html">points.h"
-
#include "publictypes.h "
-
#href_anchor"el" href="a00062_source.html">rect.h"
-
#include "scrollview.h "
+
#include <cstdio>
+
#include "unichar.h "
Go to the source code of this file.
bool divisible_blob (TBLOB *blob, bool italic_blob, TPOINT *location)&#href_anchor"memitem:a2ab5e3b619507b26900a73919b165970"> void divide_blobs (TBLOB *blob, TBLOB *other_blob, bool italic_blob, const TPOINT &location)&#href_anchor"groupheader">Macro Definition Documentation
-◆ int CompareMatchResults _ARGS ((MATCH_RESULT *Result1, MATCH_RESULT *Result2))&#href_anchor"memitem:a1b7bb9c3da4cfd66401e54922efc14f8"> void PrintMatchResult _ARGS ((FILE *File, MATCH_RESULT *MatchResult))&#href_anchor"memitem:ad7436ff9e9c8ded790312d3c8e71e975"> void PrintMatchResults _ARGS ((FILE *File, int N, MATCH_RESULT MatchResults[]))&#href_anchor"groupheader">Macro Definition Documentation
+◆
- #define EDGEPTFLAGS 4 /*concavity,length etc. */
+ #define _ARGS
+ (
+
+ s )
+ ()
-
-◆ ◆
+
all feature matchers that are to be used with the high level classifier must support the following interface. The names will, of course, be unique for each different matcher. Note also that FEATURE_STRUCT is a data structure that is defined specifically for each feature extractor/matcher pair.
-
Definition at line 97 of file blobs.h .
+
Definition at line 89 of file matchdefs.h .
-
-◆ ◆
- void divide_blobs
+ #define IsValidProto
(
- TBLOB *&#href_anchor"paramname">blob ,
+ Pid )
+ ((Pid) >= 0)
+
+
+
+◆ ◆
+
+
- tesseract: src/ccstruct/blobs.h Source File
+tesseract: src/dict/matchdefs.h Source File
-
Go to the documentation of this file.
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
+
Go to the documentation of this file.
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
18 #href_anchor"line"> 19 #define MATCHDEFS_H
-
21 #href_anchor"line"> 22 #define BLOBS_H
-
-
-
-
-
-
-
-
30 #href_anchor"code" href="a00758.html">genericvector.h"
-
-
32 #href_anchor"code" href="a00158.html">points.h"
-
-
34 #href_anchor"code" href="a00062.html">rect.h"
-
+
+
+
+
+
+
+
30 #define MAX_NUM_CLASSES INT16_MAX
+
31 #define MAX_CLASS_ID (MAX_NUM_CLASSES - 1)
+
+
+
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
53 TPOINT (int16_t vx, int16_t vy) :
x (vx),
y (vy) {}
-
-
-
-
-
-
-
-
-
-
-
-
65 return x == other.
x &&
y == other.
y ;
-
-
-
-
-
+
+
+
+
+
+
48 #define NOISE_FEATURE 254
+
49 #define MISSING_PROTO 254
+
50 #define MAX_NUM_FEAT 40
+
51 #define MAX_FEATURE_ID 250
+
+
+
+
+
+
+
+
+
+
+
+
-
-
-
-
-
-
-
-
-
80 return x * other.
y -
y * other.
x ;
-
-
-
-
-
85 return x * other.
x +
y * other.
y ;
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
125 return x_dist * x_dist * x_factor + y_dist * y_dist;
-
-
-
-
-
-
-
/usr/share/doc/packages/tesseract-ocr-devel/html/a00071.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00071.html 2022-03-29 06:12:59.485584613 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00071.html 2022-03-29 06:12:59.485584613 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/split.h File Reference
+
tesseract: src/lstm/lstm.cpp File Reference
@@ -83,200 +83,90 @@
-
#include "blobs.h "
-
#href_anchor"el" href="a00626_source.html">params.h"
-
#include "rect.h "
+
#include "lstm.h "
+
#href_anchor"el" href="a00113_source.html">fullyconnected.h"
+
#include "functions.h "
+
#href_anchor"el" href="a00209_source.html">networkscratch.h"
+
#include "tprintf.h "
Go to the source code of this file.
-
-
-
◆
+
+
+
◆
- EDGEPT * make_edgept (
- int&#href_anchor"paramname">x ,
-
-
- int
- y ,
-
- EDGEPT *&#href_anchor"paramname">next ,
-
- EDGEPT *&#href_anchor"paramname">prev
-
- )
- #define END_PARALLEL_IF_OPENMP
-
Definition at line 139 of file split.cpp .
-
139 {
-
-
141
-
-
143 this_edgept->
pos .
x = x;
-
144 this_edgept->
pos .
y = y;
-
145
-
-
147 if (prev_ol !=
nullptr && prev->
next == next) {
-
148
-
-
-
151 double cut_fraction = target_vec.length() / segment_vec.length();
-
152
-
-
-
-
-
157 ICOORD step_vec = step_end - step_start;
-
158 double target_length = step_vec.
length () * cut_fraction;
-
159
-
-
-
162 double best_dist = target_length;
-
163 for (
int s = prev->
start_step ; s < end_step; ++s) {
-
164 total_step += prev_ol->
step (s % step_length);
-
165 double dist = fabs(target_length - total_step.length());
-
166 if (dist < best_dist) {
-
167 best_dist = dist;
-
168 best_step = s + 1;
-
169 }
-
170 }
-
171
-
-
173 this_edgept->
step_count = end_step - best_step;
-
174 this_edgept->
start_step = best_step % step_length;
-
-
176 } else {
-
177
-
-
-
-
181 }
-
182
-
183 this_edgept->
next = next;
-
184 this_edgept->
prev = prev;
-
185 prev->
next = this_edgept;
-
186 next->
prev = this_edgept;
-
187
-
-
-
-
-
192 return this_edgept;
-
193 }
-
-
-
-
-
-
-
-
-
-
-
-
ICOORD step(int index) const
-
ICOORD position_at_index(int index) const
-
int32_t pathlength() const
-
-
float length() const
find length
-
-
+
Definition at line 62 of file lstm.cpp .
+
-
-
◆
+
+
◆
- void remove_edgept
+ #define PARALLEL_IF_OPENMP
(
- EDGEPT *&#href_anchor"paramname">point )
+
+ __num_threads )
/usr/share/doc/packages/tesseract-ocr-devel/html/a00071.js differs (ASCII text)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00071.js 2020-03-26 00:00:00.000000000 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00071.js 2020-03-26 00:00:00.000000000 +0000
@@ -1,7 +1,8 @@
var a00071 =
[
- [ "SPLIT", "a02889.html", "a02889" ],
- [ "make_edgept", "a00071.html#a31977dcfc91554bbf9e16dc13e79f88c", null ],
- [ "remove_edgept", "a00071.html#ada6ba0f2954768ab4b7d9d2cf165dee0", null ],
- [ "wordrec_display_splits", "a00071.html#ac251ee3f4ff4145a52a8ba54a1b26743", null ]
+ [ "END_PARALLEL_IF_OPENMP", "a00071.html#a485889d32a3c44b62120ea4db9ca5dd8", null ],
+ [ "PARALLEL_IF_OPENMP", "a00071.html#a640e05199d977160f5b1adbeb0b7526a", null ],
+ [ "SECTION_IF_OPENMP", "a00071.html#a928a2b0f3f970a1c396c0789bdd19c74", null ],
+ [ "kErrClip", "a00071.html#a20a569eeb48232f976204ba74145f99f", null ],
+ [ "kStateClip", "a00071.html#a26d5ed25b47e7e5674b03d9c62115686", null ]
];
\ No newline at end of file
/usr/share/doc/packages/tesseract-ocr-devel/html/a00071_source.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00071_source.html 2022-03-29 06:12:59.525584766 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00071_source.html 2022-03-29 06:12:59.549584857 +0000
@@ -5,7 +5,7 @@
tesseract: src/ccstruct/split.h Source File
+tesseract: src/lstm/lstm.cpp Source File
-
Go to the documentation of this file.
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
21 #href_anchor"line"> 22 #define SPLIT_H
-
-
-
-
-
-
-
29 #href_anchor"code" href="a00626.html">params.h"
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
75 float FullPriority (
int xmin,
int xmax,
double overlap_knob,
-
76 int centered_maxwidth,
double center_knob,
-
77 double width_change_knob)
const ;
-
-
-
80 bool IsHealthy (
const TBLOB & blob,
int min_points,
int min_area)
const ;
-
-
-
-
-
-
86 #ifndef GRAPHICS_DISABLED
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
120 #href_anchor"ttc" id="aa00062_html">
-
-
EDGEPT * make_edgept(int x, int y, EDGEPT *next, EDGEPT *prev)
-
bool wordrec_display_splits
-
void remove_edgept(EDGEPT *point)
-
-
#define BOOL_VAR_H(name, val, comment)
-
-
TBOX SegmentBox(const EDGEPT *end) const
-
bool EqualPos(const EDGEPT &other) const
-
-
-
bool Contains(const TPOINT &pt) const
-
-
bool Contains(const TPOINT &pt) const
-
-
-
float FullPriority(int xmin, int xmax, double overlap_knob, int centered_maxwidth, double center_knob, double width_change_knob) const
-
bool IsLittleChunk(int min_points, int min_area) const
-
-
bool ContainedByBlob(const TBLOB &blob) const
-
SPLIT(EDGEPT *pt1, EDGEPT *pt2)
-
bool IsHealthy(const TBLOB &blob, int min_points, int min_area) const
-
-
-
-
void SplitOutlineList(TESSLINE *outlines) const
-
-
TBOX bounding_box() const
-
-
-
bool ContainedByOutline(const TESSLINE &outline) const
-
-
void SplitOutline() const
-
void UnsplitOutlineList(TBLOB *blob) const
-
bool SharesPosition(const SPLIT &other) const
-
void Mark(ScrollView *window) const
-
void UnsplitOutlines() const
-
bool UsesPoint(const EDGEPT *point) const
-
+
Go to the documentation of this file.
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
19 #href_anchor"code" href="a00161.html">lstm.h"
+
+
+
+
+
+
/usr/share/doc/packages/tesseract-ocr-devel/html/a00074.html differs (HTML document, ASCII text)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00074.html 2022-03-29 06:12:59.573584950 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00074.html 2022-03-29 06:12:59.573584950 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/dppoint.h File Reference
+
tesseract: src/lstm/recodebeam.h File Reference
@@ -84,24 +84,58 @@
+
-
#include <cstdint>
-
Go to the source code of this file.
+
#include "dawg.h "
+
#href_anchor"el" href="a00062_source.html">dict.h"
+
#include "genericheap.h "
+
#href_anchor"el" href="a00530_source.html">kdpair.h"
+
#include "networkio.h "
+
#href_anchor"el" href="a00809_source.html">ratngs.h"
+
#include "unicharcompress.h "
+
#href_anchor"a00074_source.html">Go to the source code of this file.
+
+
+
+
Missing in old package: /usr/share/doc/packages/tesseract-ocr-devel/html/a00074.js
Missing in old package: /usr/share/doc/packages/tesseract-ocr-devel/html/a00074.js
/usr/share/doc/packages/tesseract-ocr-devel/html/a00074_source.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00074_source.html 2022-03-29 06:12:59.609585087 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00074_source.html 2022-03-29 06:12:59.625585149 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/dppoint.h Source File
+
tesseract: src/lstm/recodebeam.h Source File
@@ -82,128 +82,486 @@
-
Go to the documentation of this file.
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
20 #href_anchor"line"> 21 #define TESSERACT_CCSTRUCT_DPPOINT_H_
+
Go to the documentation of this file.
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
20 #href_anchor"line"> 21 #define THIRD_PARTY_TESSERACT_LSTM_RECODEBEAM_H_
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
53 total_cost_(INT32_MAX),
-
-
-
-
-
-
-
-
-
-
-
-
65 static DPPoint *
Solve (
int min_step,
int max_step,
bool debug,
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
82 local_cost_ += new_cost;
-
-
-
-
-
-
-
89 void UpdateIfBetter(int64_t cost, int32_t steps,
const DPPoint * prev,
-
90 int32_t n, int32_t sig_x, int64_t sig_xsq);
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
105 #href_anchor"comment">// TESSERACT_CCSTRUCT_DPPOINT_H_
-
-
-
-
void AddLocalCost(int new_cost)
-
static DPPoint * Solve(int min_step, int max_step, bool debug, CostFunc cost_func, int size, DPPoint *points)
-
const DPPoint * best_prev() const
-
-
int64_t CostWithVariance(const DPPoint *prev)
-
-
int64_t(DPPoint::*)(const DPPoint *) CostFunc
+
+
24 #href_anchor"code" href="a00062.html">dict.h"
+
+
26 #href_anchor"code" href="a00530.html">kdpair.h"
+
+
28 #href_anchor"code" href="a00809.html">ratngs.h"
+
+
30 #href_anchor"line"> 31 #include <set>
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
/usr/share/doc/packages/tesseract-ocr-devel/html/a00077.html differs (HTML document, ASCII text)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00077.html 2022-03-29 06:12:59.645585225 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00077.html 2022-03-29 06:12:59.649585240 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/ocrpara.h File Reference
+
tesseract: src/lstm/parallel.h File Reference
@@ -83,23 +83,25 @@
+
-
#include "publictypes.h "
-
#href_anchor"el" href="a00665_source.html">elst.h"
-
#include "strngs.h "
+
Go to the source code of this file.
struct &#href_anchor"memItemRight" valign="bottom">PARA &#href_anchor"memitem:"> class ParagraphModel class &#href_anchor"memItemRight" valign="bottom">tesseract::Parallel &#href_anchor"memberdecls">
+ namespace tesseract &#href_anchor"nav-path" class="navpath">
/usr/share/doc/packages/tesseract-ocr-devel/html/a00077_source.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00077_source.html 2022-03-29 06:12:59.673585333 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00077_source.html 2022-03-29 06:12:59.673585333 +0000
@@ -5,7 +5,7 @@
tesseract: src/ccstruct/ocrpara.h Source File
+tesseract: src/lstm/parallel.h Source File
Go to the documentation of this file.
-
-
-
-
+
+
+
+
-
+
@@ -101,213 +101,107 @@
-
-
-
20 #href_anchor"line"> 21 #define TESSERACT_CCSTRUCT_OCRPARA_H_
-
-
-
24 #href_anchor"code" href="a00665.html">elst.h"
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
121 : justification_(justification),
-
-
123 first_indent_(first_indent),
-
124 body_indent_(body_indent),
-
125 tolerance_(tolerance) {
-
-
127 int added_margin = first_indent;
-
128 if (body_indent < added_margin)
-
129 added_margin = body_indent;
-
130 margin_ += added_margin;
-
131 first_indent_ -= added_margin;
-
132 body_indent_ -= added_margin;
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
158 bool ValidFirstLine (
int lmargin,
int lindent,
int rindent,
int rmargin)
const ;
-
-
-
-
162 bool ValidBodyLine (
int lmargin,
int lindent,
int rindent,
int rmargin)
const ;
-
-
-
165 return justification_;
-
-
-
-
-
-
-
-
-
174 abs(first_indent_ - body_indent_) <= tolerance_;
-
-
-
-
-
-
/usr/share/doc/packages/tesseract-ocr-devel/html/a00080.html differs (HTML document, ASCII text)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00080.html 2022-03-29 06:12:59.693585408 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00080.html 2022-03-29 06:12:59.689585393 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/crakedge.h File Reference
+
tesseract: src/lstm/weightmatrix.cpp File Reference
@@ -83,21 +83,44 @@
+
-
#include "points.h "
-
#href_anchor"el" href="a00050_source.html">mod128.h"
+
#include "weightmatrix.h "
+
#href_anchor"el" href="a00605_source.html">intsimdmatrix.h"
+
#include "simddetect.h "
+
#href_anchor"el" href="a00827_source.html">statistc.h"
+
#include "tprintf.h "
Go to the source code of this file.
+
+
+
+
Missing in old package: /usr/share/doc/packages/tesseract-ocr-devel/html/a00080.js
Missing in old package: /usr/share/doc/packages/tesseract-ocr-devel/html/a00080.js
/usr/share/doc/packages/tesseract-ocr-devel/html/a00080_source.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00080_source.html 2022-03-29 06:12:59.717585501 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00080_source.html 2022-03-29 06:12:59.733585561 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/crakedge.h Source File
+
tesseract: src/lstm/weightmatrix.cpp Source File
@@ -82,62 +82,490 @@
-
Go to the documentation of this file.
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
+
Go to the documentation of this file.
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
18 #href_anchor"code" href="a00095.html">weightmatrix.h"
-
20 #href_anchor"line"> 21 #define CRAKEDGE_H
-
-
-
24 #href_anchor"code" href="a00050.html">mod128.h"
+
+
+
22 #href_anchor"code" href="a00590.html">simddetect.h"
+
+
24 #href_anchor"code" href="a00398.html">tprintf.h"
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
+
+
+
28 #href_anchor"line"> 29 static inline double log2(double n) {
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
49 bool skip_bias_back,
const double * u,
+
+
51 int num_results = w.
dim1 () - skip_bias_back;
+
52 int extent = w.
dim2 () - add_bias_fwd;
+
53 for (
int i = 0; i < num_results; ++i) {
+
54 const double * wi = w[i];
+
+
56 if (add_bias_fwd) total += wi[extent];
+
+
+
+
+
+
+
63 int width = input.
dim1 ();
+
64 int num_features = input.
dim2 ();
+
+
66 for (
int t = 0; t < width; ++t)
WriteStrided (t, input[t]);
+
+
+
+
+
+
+
+
+
+
+
77 float weight_range,
TRand * randomizer) {
+
+
+
80 if (randomizer !=
nullptr ) {
+
81 for (
int i = 0; i < no; ++i) {
+
82 for (
int j = 0; j < ni; ++j) {
+
83 wf_[i][j] = randomizer->
SignedRand (weight_range);
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
99 int old_no = wf_.
dim1 ();
+
100 int new_no = code_map.size();
+
+
102 std::vector<double> means(ni, 0.0);
+
103 for (
int c = 0; c < old_no; ++c) {
+
104 const double * weights = wf_[c];
+
105 for (
int i = 0; i < ni; ++i) means[i] += weights[i];
+
+
107 for (
double & mean : means) mean /= old_no;
+
+
+
+
111 int src = code_map[
dest ];
+
112 const double * src_data = src >= 0 ? old_wf[src] : means.data();
+
113 memcpy(wf_[
dest ], src_data, ni *
sizeof (*src_data));
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
128 int dim2 = wi_.
dim2 ();
+
129 for (
int t = 0; t < wi_.
dim1 (); ++t) {
+
130 double * f_line = wf_[t];
+
131 int8_t* i_line = wi_[t];
+
132 double max_abs = 0.0;
+
133 for (
int f = 0; f < dim2; ++f) {
+
134 double abs_val = fabs(f_line[f]);
+
135 if (abs_val > max_abs) max_abs = abs_val;
+
+
137 double scale = max_abs / INT8_MAX;
+
/usr/share/doc/packages/tesseract-ocr-devel/html/a00083.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00083.html 2022-03-29 06:12:59.749585623 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00083.html 2022-03-29 06:12:59.753585638 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/seam.h File Reference
+
tesseract: src/lstm/network.h File Reference
@@ -84,105 +84,89 @@
+
-
#include "blobs.h "
-
#href_anchor"el" href="a00071_source.html">split.h"
+
#include <cstdio>
+
#include <cmath>
+
#include "genericvector.h "
+
#href_anchor"el" href="a00512_source.html">helpers.h"
+
#include "matrix.h "
+
#href_anchor"el" href="a00086_source.html">networkio.h"
+
#include "serialis.h "
+
#href_anchor"el" href="a00197_source.html">static_shape.h"
+
#include "strngs.h "
+
#href_anchor"el" href="a00398_source.html">tprintf.h"
Go to the source code of this file.
class &#href_anchor"memItemRight" valign="bottom">SEAM class &#href_anchor"memItemRight" valign="bottom">tesseract::Network &#href_anchor"memberdecls">
- using PRIORITY = float
-void start_seam_list (TWERD *word, GenericVector < SEAM * > *seam_array)&#href_anchor"groupheader">Typedef Documentation
-◆
-
-
-
Definition at line 36 of file seam.h .
-
-
-
◆
-
-
-
- void start_seam_list
- (
- TWERD *&#href_anchor"paramname">word ,
-
- GenericVector < SEAM * > *&#href_anchor"paramname">seam_array
-
- )
-
-
-
-
-
Definition at line 263 of file seam.cpp .
-
263 {
-
-
-
266
-
267 for (
int b = 1; b < word->
NumBlobs (); ++b) {
-
268 TBOX bbox = word->
blobs [b - 1]->bounding_box();
-
269 TBOX nbox = word->
blobs [b]->bounding_box();
-
270 location.
x = (bbox.
right () + nbox.
left ()) / 2;
-
-
-
273 }
-
274 }
-
-
-
-
-
-
-
GenericVector< TBLOB * > blobs
-
-
-
-
-
-
-
-
-
namespace tesseract &#href_anchor"memberdecls">
+ enum tesseract::NetworkType { tesseract::NT_NONE
+, tesseract::NT_INPUT
+, tesseract::NT_CONVOLVE
+, tesseract::NT_MAXPOOL
+, tesseract::NT_PARALLEL
+, tesseract::NT_REPLICATED
+, tesseract::NT_PAR_RL_LSTM
+, tesseract::NT_PAR_UD_LSTM
+, tesseract::NT_PAR_2D_LSTM
+, tesseract::NT_SERIES
+, tesseract::NT_RECONFIG
+, tesseract::NT_XREVERSED
+, tesseract::NT_YREVERSED
+, tesseract::NT_XYTRANSPOSE
+, tesseract::NT_LSTM
+, tesseract::NT_LSTM_SUMMARY
+, tesseract::NT_LOGISTIC
+, tesseract::NT_POSCLIP
+, tesseract::NT_SYMCLIP
+, tesseract::NT_TANH
+, tesseract::NT_RELU
+, tesseract::NT_LINEAR
+, tesseract::NT_SOFTMAX
+, tesseract::NT_SOFTMAX_NO_CTC
+, tesseract::NT_LSTM_SOFTMAX
+, tesseract::NT_LSTM_SOFTMAX_ENCODED
+, tesseract::NT_TENSORFLOW
+, tesseract::NT_COUNT
+ enum tesseract::NetworkFlags { tesseract::NF_LAYER_SPECIFIC_LR = 64
+, tesseract::NF_ADAM = 128
+ } enum tesseract::TrainingState { tesseract::TS_DISABLED
+, tesseract::TS_ENABLED
+, tesseract::TS_TEMP_DISABLE
+, tesseract::TS_RE_ENABLE
+ }
/usr/share/doc/packages/tesseract-ocr-devel/html/a00083.js differs (ASCII text)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00083.js 2020-03-26 00:00:00.000000000 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00083.js 2020-03-26 00:00:00.000000000 +0000
@@ -1,6 +1,44 @@
var a00083 =
[
- [ "SEAM", "a02885.html", "a02885" ],
- [ "PRIORITY", "a00083.html#a64c977704107e265e5997b915a83ccf2", null ],
- [ "start_seam_list", "a00083.html#a7a7eb511f9a83da65455fe0a2668bc6b", null ]
+ [ "tesseract::Network", "a03705.html", "a03705" ],
+ [ "NetworkFlags", "a00083.html#aaca6aa7e71197530dd154e11138da6e8", [
+ [ "NF_LAYER_SPECIFIC_LR", "a00083.html#aaca6aa7e71197530dd154e11138da6e8a575556edf34617c58c34920f0190b85f", null ],
+ [ "NF_ADAM", "a00083.html#aaca6aa7e71197530dd154e11138da6e8a98ee63d76d379ea872c9d6f055ad3ba2", null ]
+ ] ],
+ [ "NetworkType", "a00083.html#a9c90202f597e58fba8911e0d29d9b198", [
+ [ "NT_NONE", "a00083.html#a9c90202f597e58fba8911e0d29d9b198a3eb5e1d7abaa3d892c82af90ba2df0fe", null ],
+ [ "NT_INPUT", "a00083.html#a9c90202f597e58fba8911e0d29d9b198a765ebca107d47f7fadfe383b13d4d872", null ],
+ [ "NT_CONVOLVE", "a00083.html#a9c90202f597e58fba8911e0d29d9b198a34b034cd5d766578ff90dcb664cf4043", null ],
+ [ "NT_MAXPOOL", "a00083.html#a9c90202f597e58fba8911e0d29d9b198a1cb4804f20581fcf9f062371ddc33c3a", null ],
+ [ "NT_PARALLEL", "a00083.html#a9c90202f597e58fba8911e0d29d9b198a57231f76d0f3fa44273dbacbdfcbafb9", null ],
+ [ "NT_REPLICATED", "a00083.html#a9c90202f597e58fba8911e0d29d9b198ae9fa95fb35d7696f132f6fd76073ded4", null ],
+ [ "NT_PAR_RL_LSTM", "a00083.html#a9c90202f597e58fba8911e0d29d9b198adbe09d99902b49d235ca444a6b0d0682", null ],
+ [ "NT_PAR_UD_LSTM", "a00083.html#a9c90202f597e58fba8911e0d29d9b198a54e083558478b1c017ae44293b76acd5", null ],
+ [ "NT_PAR_2D_LSTM", "a00083.html#a9c90202f597e58fba8911e0d29d9b198a5d30bba16887f026be0f429e06f6095e", null ],
+ [ "NT_SERIES", "a00083.html#a9c90202f597e58fba8911e0d29d9b198aa8f65aa7ff6e3300e56389aa28c52f04", null ],
+ [ "NT_RECONFIG", "a00083.html#a9c90202f597e58fba8911e0d29d9b198a739ff05696ee5a444094e55a9bbde2b3", null ],
+ [ "NT_XREVERSED", "a00083.html#a9c90202f597e58fba8911e0d29d9b198a23eb489605474a1009d27eafaa99faae", null ],
+ [ "NT_YREVERSED", "a00083.html#a9c90202f597e58fba8911e0d29d9b198a631f64444e00eff9f5bcfa180bf01792", null ],
+ [ "NT_XYTRANSPOSE", "a00083.html#a9c90202f597e58fba8911e0d29d9b198aa3ce4edd0778adde37a2c476d38b11b1", null ],
+ [ "NT_LSTM", "a00083.html#a9c90202f597e58fba8911e0d29d9b198a2dd05ea92fab5779a1c1b512795eb097", null ],
+ [ "NT_LSTM_SUMMARY", "a00083.html#a9c90202f597e58fba8911e0d29d9b198a61ac09a484fbadf9dbbeb9e10df26c57", null ],
+ [ "NT_LOGISTIC", "a00083.html#a9c90202f597e58fba8911e0d29d9b198a434b4a88c75870ee1cdc320536383fae", null ],
+ [ "NT_POSCLIP", "a00083.html#a9c90202f597e58fba8911e0d29d9b198a9474b507410af4e788557f1ff2b42674", null ],
+ [ "NT_SYMCLIP", "a00083.html#a9c90202f597e58fba8911e0d29d9b198a5cbeafa21116987ff9caae39e9c07c42", null ],
+ [ "NT_TANH", "a00083.html#a9c90202f597e58fba8911e0d29d9b198ad501a0811a40ec36658696e6b76f6334", null ],
+ [ "NT_RELU", "a00083.html#a9c90202f597e58fba8911e0d29d9b198a22a33a868490aea6d2c5366805ae5076", null ],
+ [ "NT_LINEAR", "a00083.html#a9c90202f597e58fba8911e0d29d9b198a0c1254967e2d5a03af5c52c7de9eaac6", null ],
+ [ "NT_SOFTMAX", "a00083.html#a9c90202f597e58fba8911e0d29d9b198a3949f2865c2ab2dd533f6d26065b32c8", null ],
+ [ "NT_SOFTMAX_NO_CTC", "a00083.html#a9c90202f597e58fba8911e0d29d9b198ad12de7ea44323bead2977fed1857b84e", null ],
+ [ "NT_LSTM_SOFTMAX", "a00083.html#a9c90202f597e58fba8911e0d29d9b198a9a70fad2938bf40bdce9de74aa531a5b", null ],
+ [ "NT_LSTM_SOFTMAX_ENCODED", "a00083.html#a9c90202f597e58fba8911e0d29d9b198a551d4e057f29f039f2468de6e2ad9bf9", null ],
+ [ "NT_TENSORFLOW", "a00083.html#a9c90202f597e58fba8911e0d29d9b198a83ad2fcd7f3fb898d1e5cbc987f472fd", null ],
+ [ "NT_COUNT", "a00083.html#a9c90202f597e58fba8911e0d29d9b198ae51d62ce0cecb096ad08960d9108aa80", null ]
+ ] ],
+ [ "TrainingState", "a00083.html#a9855a974d6a4ed30fa0a4a4164639989", [
+ [ "TS_DISABLED", "a00083.html#a9855a974d6a4ed30fa0a4a4164639989a7dcb0f12f914897aab3df6760bb0ef11", null ],
+ [ "TS_ENABLED", "a00083.html#a9855a974d6a4ed30fa0a4a4164639989a6748e3864f889f02ae7f65f53af125e3", null ],
+ [ "TS_TEMP_DISABLE", "a00083.html#a9855a974d6a4ed30fa0a4a4164639989a5f4e47794c37db71080e921c317a90c9", null ],
+ [ "TS_RE_ENABLE", "a00083.html#a9855a974d6a4ed30fa0a4a4164639989aa9a24b714a92d4f8955c6ce122cdc37e", null ]
+ ] ]
];
\ No newline at end of file
/usr/share/doc/packages/tesseract-ocr-devel/html/a00083_source.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00083_source.html 2022-03-29 06:12:59.789585776 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00083_source.html 2022-03-29 06:12:59.801585822 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/seam.h Source File
+
tesseract: src/lstm/network.h Source File
@@ -82,253 +82,427 @@
-
Go to the documentation of this file.
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
19 #href_anchor"line"> 20 #define SEAM_H
-
-
-
-
24 #include "config_auto.h"
-
-
-
-
-
-
-
31 #href_anchor"code" href="a00071.html">split.h"
+
Go to the documentation of this file.
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
18 #href_anchor"line"> 19 #define TESSERACT_LSTM_NETWORK_H_
+
+
+
+
+
+
25 #href_anchor"code" href="a00512.html">helpers.h"
+
+
27 #href_anchor"code" href="a00086.html">networkio.h"
+
+
29 #href_anchor"code" href="a00197.html">static_shape.h"
+
+
31 #href_anchor"code" href="a00398.html">tprintf.h"
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
68 float max_total_priority)
const ;
-
-
-
-
-
-
74 for (
int s = 0; s < num_splits_; ++s) {
-
-
-
-
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
-
-
-
-
83 for (
int s = 0; s < num_splits_; ++s) {
-
84 if (splits_[s].
UsesPoint (point))
return true ;
-
-
-
-
-
-
90 for (
int s = 0; s < num_splits_; ++s) {
-
91 for (
int t = 0; t < other.num_splits_; ++t)
-
-
-
-
-
-
-
98 for (
int s = 0; s < num_splits_; ++s) {
-
-
100 for (
int t = 0; t < other.num_splits_; ++t) {
-
-
102 if (split1_box.
y_overlap (split2_box))
return true ;
-
-
-
-
-
/usr/share/doc/packages/tesseract-ocr-devel/html/a00086.html differs (HTML document, ASCII text)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00086.html 2022-03-29 06:12:59.817585882 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00086.html 2022-03-29 06:12:59.817585882 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/stepblob.cpp File Reference
+
tesseract: src/lstm/networkio.h File Reference
@@ -83,44 +83,32 @@
+
-
#include "stepblob.h "
-
#href_anchor"el" href="a00758_source.html">genericvector.h"
-
#include "points.h "
+
#include <cmath>
+
#include <cstdio>
+
#include <vector>
+
#include "genericvector.h "
+
#href_anchor"el" href="a00512_source.html">helpers.h"
+
#include "static_shape.h "
+
#href_anchor"el" href="a00200_source.html">stridemap.h"
+
#include "weightmatrix.h "
Go to the source code of this file.
-
-
-
◆
-
-
-
-
-
- const double kMaxPerimeterWidthRatio = 8.0
-
-
-
-
-
-
-
-
+
+
class &#href_anchor"memItemRight" valign="bottom">tesseract::NetworkIO
+
&#href_anchor"memberdecls">
+
+
namespace tesseract
+
&#href_anchor"nav-path" class="navpath">
/usr/share/doc/packages/tesseract-ocr-devel/html/a00086_source.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00086_source.html 2022-03-29 06:12:59.857586035 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00086_source.html 2022-03-29 06:12:59.861586051 +0000
@@ -5,7 +5,7 @@
tesseract: src/ccstruct/stepblob.cpp Source File
+tesseract: src/lstm/networkio.h Source File
-
Go to the documentation of this file.
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
20 #href_anchor"line"> 21 #include "config_auto.h"
-
-
-
-
25 #href_anchor"comment">// for pixCreate, pixGetDepth
-
-
27 #href_anchor"code" href="a00158.html">points.h"
-
-
-
-
-
+
Go to the documentation of this file.
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
19 #href_anchor"line"> 20 #define TESSERACT_LSTM_NETWORKIO_H_
+
+
+
+
+
+
+
27 #href_anchor"code" href="a00512.html">helpers.h"
+
+
29 #href_anchor"code" href="a00200.html">stridemap.h"
+
+
+
-
-
-
-
-
-
-
-
41 static void position_outline(
-
-
43 C_OUTLINE_LIST *destlist
-
-
-
46 C_OUTLINE_IT it = destlist;
-
-
48 C_OUTLINE_IT child_it = outline->
child ();
-
-
-
-
52 dest_outline = it.data ();
-
-
54 if (*dest_outline < *outline) {
-
-
56 dest_outline = it.extract ();
-
-
58 it.add_after_then_move (outline);
-
-
60 child_it.add_to_end (dest_outline);
-
61 while (!it.at_last ()) {
-
-
-
64 dest_outline = it.data ();
-
65 if (*dest_outline < *outline) {
-
-
67 dest_outline = it.extract ();
-
68 child_it.add_to_end (dest_outline);
-
-
-
-
-
-
-
-
-
77 else if (*outline < *dest_outline) {
-
78 position_outline (outline, dest_outline->
child ());
-
-
-
-
-
-
84 while (!it.at_first ());
-
-
86 it.add_to_end (outline);
-
-
-
-
-
-
-
-
-
-
-
97 #ifndef GRAPHICS_DISABLED
-
98 static void plot_outline_list(
-
-
-
-
-
-
-
105 C_OUTLINE_IT it = list;
-
-
107 for (it.mark_cycle_pt (); !it.cycled_list (); it.forward ()) {
-
108 outline = it.data ();
-
-
110 outline->
plot (window, colour);
-
111 if (!outline->
child ()->empty ())
-
112 plot_outline_list (outline->
child (), window,
-
113 child_colour, child_colour);
-
-
-
-
-
-
119 static void plot_normed_outline_list(
const DENORM & denorm,
-
120 C_OUTLINE_LIST *list,
-
-
-
-
124 C_OUTLINE_IT it(list);
-
125 for (it.mark_cycle_pt(); !it.cycled_list(); it.forward()) {
-
-
-
128 if (!outline->
child ()->empty())
-
129 plot_normed_outline_list(denorm, outline->
child (), child_colour,
-
130 child_colour, window);
-
-
-
-
-
-
-
-
-
-
+
+
+
+
+
+
+
+
+
+
+
+
/usr/share/doc/packages/tesseract-ocr-devel/html/a00089.html differs (HTML document, ASCII text)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00089.html 2022-03-29 06:12:59.881586127 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00089.html 2022-03-29 06:12:59.881586127 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/stepblob.h File Reference
+
tesseract: src/lstm/reversed.h File Reference
@@ -83,25 +83,26 @@
+
-
#include <cstdint>
-
#include "coutln.h "
-
#href_anchor"el" href="a00665_source.html">elst.h"
-
#include "points.h "
-
#href_anchor"el" href="a00062_source.html">rect.h"
-
#include "scrollview.h "
+
#include "matrix.h "
+
#href_anchor"el" href="a00188_source.html">plumbing.h"
Go to the source code of this file.
class &#href_anchor"memItemRight" valign="bottom">C_BLOB class &#href_anchor"memItemRight" valign="bottom">tesseract::Reversed &#href_anchor"memberdecls">
+ namespace tesseract &#href_anchor"nav-path" class="navpath">
/usr/share/doc/packages/tesseract-ocr-devel/html/a00089_source.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00089_source.html 2022-03-29 06:12:59.901586203 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00089_source.html 2022-03-29 06:12:59.909586234 +0000
@@ -5,7 +5,7 @@
tesseract: src/ccstruct/stepblob.h Source File
+tesseract: src/lstm/reversed.h Source File
-
Go to the documentation of this file.
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
20 #href_anchor"line"> 21 #define STEPBLOB_H
-
-
-
-
25 #href_anchor"code" href="a00665.html">elst.h"
-
-
27 #href_anchor"code" href="a00062.html">rect.h"
-
-
-
-
+
Go to the documentation of this file.
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
19 #href_anchor"line"> 20 #define TESSERACT_LSTM_REVERSED_H_
+
+
+
23 #href_anchor"code" href="a00188.html">plumbing.h"
+
+
+
+
+
+
+
+
-
-
-
+
+
+
-
-
-
-
-
41 explicit C_BLOB (C_OUTLINE_LIST *outline_list);
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
56 static void ConstructBlobsFromOutlines(
bool good_blob,
-
57 C_OUTLINE_LIST* outline_list,
-
58 C_BLOB_IT* good_blobs_it,
-
59 C_BLOB_IT* bad_blobs_it);
-
-
-
-
-
64 void CheckInverseFlagAndDirection();
-
-
-
-
+
+
+
+
+
+
+
+
+
+
+
47 if (net_spec[0] ==
'L' ) {
+
+
+
+
+
+
+
+
+
+
+
58 for (
int i = 0; i < net_spec.
length (); ++i) {
+
59 if (net_spec[i] == from) net_spec[i] = to;
+
+
+
+
+
+
+
+
+
-
-
-
-
-
74 TBOX bounding_box()
const ;
-
-
-
-
78 int32_t count_transitions(
-
-
-
81 void move(
const ICOORD vec);
-
82 void rotate(
const FCOORD & rotation);
-
-
-
-
86 void ComputeEdgeOffsets(
int threshold, Pix* pix);
-
-
-
-
90 int16_t EstimateBaselinePosition();
-
-
-
-
-
-
-
-
98 #ifndef GRAPHICS_DISABLED
-
-
-
-
-
-
-
-
106 void plot_normed(
const DENORM & denorm,
-
-
-
-
110 #href_anchor"comment">// GRAPHICS_DISABLED
-
-
-
113 if (!outlines.empty ())
-
-
-
-
-
/usr/share/doc/packages/tesseract-ocr-devel/html/a00092.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00092.html 2022-03-29 06:12:59.929586311 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00092.html 2022-03-29 06:12:59.929586311 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/otsuthr.cpp File Reference
+
tesseract: src/lstm/reconfig.h File Reference
@@ -83,35 +83,27 @@
+
-
#include "otsuthr.h "
-
#href_anchor"el" href="a00635_source.html">helpers.h"
+
#include "genericvector.h "
+
#href_anchor"el" href="a00725_source.html">matrix.h"
+
#include "network.h "
Go to the source code of this file.
+class &#href_anchor"memItemRight" valign="bottom">tesseract::Reconfig &#href_anchor"memberdecls">
namespace &#href_anchor"memItemRight" valign="bottom">tesseract &#href_anchor"memberdecls">
- int tesseract::OtsuThreshold (Pix *src_pix, int left, int top, int width, int height, int **thresholds, int **hi_values) void tesseract::HistogramRect (Pix *src_pix, int channel, int left, int top, int width, int height, int *histogram) int tesseract::OtsuStats (const int *histogram, int *H_out, int *omega0_out)
-
-
-
-
+
namespace tesseract
+
&#href_anchor"nav-path" class="navpath">
/usr/share/doc/packages/tesseract-ocr-devel/html/a00092_source.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00092_source.html 2022-03-29 06:12:59.945586372 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00092_source.html 2022-03-29 06:12:59.945586372 +0000
@@ -5,7 +5,7 @@
tesseract: src/ccstruct/otsuthr.cpp Source File
+tesseract: src/lstm/reconfig.h Source File
-
Go to the documentation of this file.
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
20 #href_anchor"code" href="a00203.html">otsuthr.h"
+
Go to the documentation of this file.
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
19 #href_anchor"line"> 20 #define TESSERACT_LSTM_RECONFIG_H_
-
-
23 #include "allheaders.h"
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
40 int OtsuThreshold (Pix* src_pix,
int left,
int top,
int width,
int height,
-
41 int ** thresholds,
int ** hi_values) {
-
42 int num_channels = pixGetDepth(src_pix) / 8;
-
-
-
45 int best_hi_value = 1;
-
46 int best_hi_index = 0;
-
47 bool any_good_hivalue =
false ;
-
48 double best_hi_dist = 0.0;
-
49 *thresholds =
new int [num_channels];
-
50 *hi_values =
new int [num_channels];
-
-
-
-
-
55 int * histogramAllChannels =
new int [
kHistogramSize * num_channels];
-
-
-
-
59 if (od.selectedDeviceIsOpenCL() && (num_channels == 1 || num_channels == 4) &&
-
60 top == 0 && left == 0) {
-
61 od.HistogramRectOCL(pixGetData(src_pix), num_channels,
-
62 pixGetWpl(src_pix) * 4, left, top, width, height,
-
-
-
-
66 for (
int ch = 0; ch < num_channels; ++ch) {
-
67 (*thresholds)[ch] = -1;
-
68 (*hi_values)[ch] = -1;
-
-
-
-
72 int best_t =
OtsuStats (histogram, &H, &best_omega_0);
-
73 if (best_omega_0 == 0 || best_omega_0 == H) {
-
-
-
-
-
-
-
80 int hi_value = best_omega_0 < H * 0.5;
-
81 (*thresholds)[ch] = best_t;
-
82 if (best_omega_0 > H * 0.75) {
-
83 any_good_hivalue =
true ;
-
-
85 }
else if (best_omega_0 < H * 0.25) {
-
86 any_good_hivalue =
true ;
-
-
-
-
90 double hi_dist = hi_value ? (H - best_omega_0) : best_omega_0;
-
91 if (hi_dist > best_hi_dist) {
-
92 best_hi_dist = hi_dist;
-
93 best_hi_value = hi_value;
-
-
-
-
-
-
-
100 for (
int ch = 0; ch < num_channels; ++ch) {
-
101 (*thresholds)[ch] = -1;
-
102 (*hi_values)[ch] = -1;
-
-
-
105 HistogramRect (src_pix, ch, left, top, width, height, histogram);
-
-
-
108 int best_t =
OtsuStats (histogram, &H, &best_omega_0);
-
109 if (best_omega_0 == 0 || best_omega_0 == H) {
-
-
-
-
-
-
-
116 int hi_value = best_omega_0 < H * 0.5;
-
117 (*thresholds)[ch] = best_t;
-
118 if (best_omega_0 > H * 0.75) {
-
119 any_good_hivalue =
true ;
-
120 (*hi_values)[ch] = 0;
-
121 }
else if (best_omega_0 < H * 0.25) {
-
122 any_good_hivalue =
true ;
-
123 (*hi_values)[ch] = 1;
-
-
-
126 double hi_dist = hi_value ? (H - best_omega_0) : best_omega_0;
-
127 if (hi_dist > best_hi_dist) {
-
128 best_hi_dist = hi_dist;
-
129 best_hi_value = hi_value;
-
-
-
-
-
-
-
136 delete [] histogramAllChannels;
-
-
-
139 if (!any_good_hivalue) {
-
-
141 (*hi_values)[best_hi_index] = best_hi_value;
-
-
-
-
-
-
-
-
-
-
151 int left,
int top,
int width,
int height,
-
-
153 int num_channels = pixGetDepth(src_pix) / 8;
-
154 channel =
ClipToRange (channel, 0, num_channels - 1);
-
155 int bottom = top + height;
-
-
157 int src_wpl = pixGetWpl(src_pix);
-
158 l_uint32* srcdata = pixGetData(src_pix);
-
159 for (
int y = top; y < bottom; ++y) {
-
160 const l_uint32* linedata = srcdata + y * src_wpl;
-
161 for (
int x = 0; x < width; ++x) {
-
162 int pixel = GET_DATA_BYTE(linedata, (x + left) * num_channels + channel);
-
-
-
/usr/share/doc/packages/tesseract-ocr-devel/html/a00095.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00095.html 2022-03-29 06:12:59.969586464 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00095.html 2022-03-29 06:12:59.965586448 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/werd.h File Reference
+
tesseract: src/lstm/weightmatrix.h File Reference
@@ -84,228 +84,29 @@
+
-
#include "bits16.h "
-
#href_anchor"el" href="a00644_source.html">elst2.h"
-
#include "params.h "
-
#href_anchor"el" href="a00089_source.html">stepblob.h"
-
#include "strngs.h "
-
#href_anchor"el" href="a00119_source.html">ocrrow.h"
+
#include <memory>
+
#include "genericvector.h "
+
#href_anchor"el" href="a00605_source.html">intsimdmatrix.h"
+
#include "matrix.h "
+
#href_anchor"el" href="a00398_source.html">tprintf.h"
Go to the source code of this file.
class &#href_anchor"memItemRight" valign="bottom">WERD class &#href_anchor"memItemRight" valign="bottom">tesseract::TransposedArray &#href_anchor"memitem:"> class tesseract::WeightMatrix &#href_anchor"memberdecls">
- enum WERD_FLAGS { W_SEGMENTED
-, W_ITALIC
-, W_BOLD
-, W_BOL
-, W_EOL
-, W_NORMALIZED
-, W_SCRIPT_HAS_XHEIGHT
-, W_SCRIPT_IS_LATIN
-, W_DONT_CHOP
-, W_REP_CHAR
-, W_FUZZY_SP
-, W_FUZZY_NON
-, W_INVERSE
- enum DISPLAY_FLAGS { DF_BOX
-, DF_TEXT
-, DF_POLYGONAL
-, DF_EDGE_STEP
-, DF_BN_POLYGONAL
-, DF_BLAMER
-
-
-
-
◆
-
-
-
-
-Enumerator Bounding box.
-Correct ascii.
-Polyg approx.
-Edge steps.
-BL normalisd polyapx.
-Blamer information.
-
-
-
Definition at line 44 of file werd.h .
-
44 {
-
45
-
-
-
-
-
-
-
52 };
-
@ DF_BN_POLYGONAL
BL normalisd polyapx.
-
@ DF_EDGE_STEP
Edge steps.
-
-
-
@ DF_BLAMER
Blamer information.
-
@ DF_POLYGONAL
Polyg approx.
-
-
-
-
-
◆
-
-
-
-
-Enumerator correctly segmented
-italic text
-bold text
-start of line
-end of line
-flags
-x-height concept makes sense.
-Special case latin for y. splitting.
-fixed pitch chopped
-repeated character
-fuzzy space
-fuzzy nonspace
-white on black
-
-
-
Definition at line 28 of file werd.h .
-
28 {
-
-
-
-
-
-
-
-
-
-
-
-
-
-
42 };
-
-
-
@ W_SCRIPT_HAS_XHEIGHT
x-height concept makes sense.
-
-
@ W_FUZZY_NON
fuzzy nonspace
-
-
@ W_SEGMENTED
correctly segmented
-
@ W_SCRIPT_IS_LATIN
Special case latin for y. splitting.
-
@ W_REP_CHAR
repeated character
-
@ W_DONT_CHOP
fixed pitch chopped
-
@ W_INVERSE
white on black
-
-
-
-
-
-
/usr/share/doc/packages/tesseract-ocr-devel/html/a00095_source.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00095_source.html 2022-03-29 06:12:59.997586571 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00095_source.html 2022-03-29 06:13:00.001586586 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/werd.h Source File
+
tesseract: src/lstm/weightmatrix.h Source File
@@ -82,271 +82,241 @@
-
Go to the documentation of this file.
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
19 #href_anchor"line"> 20 #define WERD_H
-
-
-
23 #href_anchor"code" href="a00644.html">elst2.h"
-
-
25 #href_anchor"code" href="a00089.html">stepblob.h"
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
63 WERD (C_BLOB_LIST* blob_list, uint8_t blanks,
const char *
text );
-
-
-
-
-
68 WERD (C_BLOB_LIST* blob_list,
WERD * clone);
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
87 C_BLOB_LIST* orphan_blobs);
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
115 void set_text (
const char * new_text) { correct = new_text; }
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
139 #ifndef GRAPHICS_DISABLED
-
-
-
-
-
-
-
-
-
-
-
-
151 #href_anchor"comment">// GRAPHICS_DISABLED
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
171 bool * make_next_word_fuzzy);
-
-
-
-
-
-
177 int16_t script_id_ = 0;
-
-
-
180 C_BLOB_LIST rej_cblobs;
+
Go to the documentation of this file.
+
+
+
+
/usr/share/doc/packages/tesseract-ocr-devel/html/a00098.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00098.html 2022-03-29 06:13:00.021586663 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00098.html 2022-03-29 06:13:00.021586663 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/matrix.h File Reference
+
tesseract: src/lstm/convolve.h File Reference
@@ -84,79 +84,26 @@
+
-
#include <algorithm>
-
#include <cmath>
-
#include <cstdint>
-
#include <cstdio>
-
#include <cstring>
-
#include "errcode.h "
-
#href_anchor"el" href="a00635_source.html">helpers.h"
-
#include "kdpair.h "
-
#href_anchor"el" href="a00158_source.html">points.h"
-
#include "serialis.h "
+
#include "genericvector.h "
+
#href_anchor"el" href="a00725_source.html">matrix.h"
+
#include "network.h "
Go to the source code of this file.
-using MatrixCoordPair = tesseract::KDPairInc < float, MATRIX_COORD >&#href_anchor"groupheader">Macro Definition Documentation
-◆
-
-
-
- #define NOT_CLASSIFIED static_cast<BLOB_CHOICE_LIST*>(nullptr)
-
-
-
-
◆
+
+
namespace tesseract
+
&#href_anchor"nav-path" class="navpath">
/usr/share/doc/packages/tesseract-ocr-devel/html/a00098_source.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00098_source.html 2022-03-29 06:13:00.061586816 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00098_source.html 2022-03-29 06:13:00.053586785 +0000
@@ -5,7 +5,7 @@
tesseract: src/ccstruct/matrix.h Source File
+tesseract: src/lstm/convolve.h Source File
-
Go to the documentation of this file.
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
+
Go to the documentation of this file.
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
20 #href_anchor"line"> 21 #define TESSERACT_LSTM_CONVOLVE_H_
+
+
+
24 #href_anchor"code" href="a00725.html">matrix.h"
+
-
27 #href_anchor"line"> 28 #define TESSERACT_CCSTRUCT_MATRIX_H_
-
-
-
-
-
-
-
-
36 #href_anchor"code" href="a00635.html">helpers.h"
-
-
38 #href_anchor"code" href="a00158.html">points.h"
-
-
-
41 class BLOB_CHOICE_LIST;
-
-
-
44 #define NOT_CLASSIFIED static_cast<BLOB_CHOICE_LIST*>(nullptr)
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
95 int new_size = size1 * size2 + pad;
-
-
-
-
-
-
-
-
-
104 for (
int i = size1 * size2; i < new_size; ++i)
array_ [i] =
empty_ ;
-
-
-
-
108 void Resize (
int size1,
int size2,
const T& empty) {
-
-
-
-
-
-
-
-
-
117 int new_size = size1 * size2;
-
118 T* new_array =
new T[new_size];
-
119 for (
int col = 0; col < size1; ++col) {
-
120 for (
int row = 0; row < size2; ++row) {
-
121 int old_index = col *
dim2 () + row;
-
122 int new_index = col * size2 + row;
-
-
124 new_array[new_index] =
array_ [old_index];
-
-
126 new_array[new_index] =
empty_ ;
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
141 for (
int i = 0; i < total_size; ++i)
/usr/share/doc/packages/tesseract-ocr-devel/html/a00101.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00101.html 2022-03-29 06:13:00.093586938 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00101.html 2022-03-29 06:13:00.089586923 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/boxread.cpp File Reference
+
tesseract: src/lstm/maxpool.cpp File Reference
@@ -83,550 +83,20 @@
+
-
#include "boxread.h "
-
#href_anchor"el" href="a00647_source.html">errcode.h"
-
#include "fileerr.h "
-
#href_anchor"el" href="a00758_source.html">genericvector.h"
-
#include "helpers.h "
-
#href_anchor"el" href="a00062_source.html">rect.h"
-
#include "strngs.h "
-
#href_anchor"el" href="a00668_source.html">tprintf.h"
-
#include "unichar.h "
+
Go to the source code of this file.
-FILE *&#href_anchor"memItemRight" valign="bottom">OpenBoxFile (const STRING &fname) &#href_anchor"memitem:ab3161d7c887ead295045650e34d690fc"> bool ReadAllBoxes (int target_page, bool skip_blanks, const STRING &filename, GenericVector < TBOX > *boxes, GenericVector < STRING > *texts, GenericVector < STRING > *box_texts, GenericVector < int > *pages)&#href_anchor"memitem:a166c16e9d8d3b3eda52ceb38c06dce64"> bool ReadMemBoxes (int target_page, bool skip_blanks, const char *box_data, bool continue_on_failure, GenericVector < TBOX > *boxes, GenericVector < STRING > *texts, GenericVector < STRING > *box_texts, GenericVector < int > *pages)&#href_anchor"memitem:af8465e130ce34e31d743881d41ecba78"> STRING &#href_anchor"memItemRight" valign="bottom">BoxFileName (const STRING &image_filename)&#href_anchor"memitem:a8236e051549f8174e263196602edadc5"> bool ReadNextBox (int *line_number, FILE *box_file, STRING *utf8_str, TBOX *bounding_box)&#href_anchor"memitem:a566c682100d2b5f0a20f39bf99ad2d30"> bool ReadNextBox (int target_page, int *line_number, FILE *box_file, STRING *utf8_str, TBOX *bounding_box)&#href_anchor"memitem:acd464bb54397318c2f6fb8db9e644fd0"> bool ParseBoxFileStr (const char *boxfile_str, int *page_number, STRING *utf8_str, TBOX *bounding_box)&#href_anchor"memitem:a12d56f93e661ff006a382fb8d7314898"> void MakeBoxFileStr (const char *unichar_str, const TBOX &box, int page_num, STRING *box_str)&#href_anchor"groupheader">Function Documentation
-◆
-
-
-
- STRING BoxFileName (
- const STRING &&#href_anchor"paramname">image_filename )
-
-
-
-
-
Definition at line 106 of file boxread.cpp .
-
106 {
-
107 STRING box_filename = image_filename;
-
108 const char *lastdot = strrchr(box_filename.
string (),
'.' );
-
109 if (lastdot != nullptr )
-
-
111
-
112 box_filename += ".box" ;
-
113 return box_filename;
-
114 }
-
-
void truncate_at(int32_t index)
-
const char * string() const
-
-
-
◆
-
-
-
- void MakeBoxFileStr
- (
- const char *
- unichar_str ,
-
- const TBOX &&#href_anchor"paramname">box ,
-
-
- int
- page_num ,
-
- STRING *&#href_anchor"paramname">box_str
-
- )
-
-
-
-
-
Definition at line 242 of file boxread.cpp .
-
243 {
-
244 *box_str = unichar_str;
-
-
-
-
-
-
250 }
-
-
-
-
-
void add_str_int(const char *str, int number)
-
-
-
◆
-
-
-
- FILE * OpenBoxFile
- (
- const STRING &&#href_anchor"paramname">fname )
-
-
-
-
-
Definition at line 36 of file boxread.cpp .
-
36 {
-
-
38 FILE* box_file = nullptr ;
-
39 if (!(box_file = fopen(filename.
string (),
"rb" ))) {
-
-
-
42 }
-
43 return box_file;
-
44 }
-
STRING BoxFileName(const STRING &image_filename)
-
-
constexpr ERRCODE CANTOPENFILE("Can't open file")
-
void error(const char *caller, TessErrorLogCode action, const char *format,...) const
-
-
-
◆
-
-
-
- bool ParseBoxFileStr
- (
- const char *
- boxfile_str ,
-
- int *
- page_number ,
-
- STRING *&#href_anchor"paramname">utf8_str ,
-
- TBOX *&#href_anchor"paramname">bounding_box
-
- )
-
-
-
-
-
Definition at line 174 of file boxread.cpp .
-
175 {
-
176 *bounding_box =
TBOX ();
/usr/share/doc/packages/tesseract-ocr-devel/html/a00101_source.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00101_source.html 2022-03-29 06:13:00.121587045 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00101_source.html 2022-03-29 06:13:00.125587060 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/boxread.cpp Source File
+
tesseract: src/lstm/maxpool.cpp Source File
@@ -82,306 +82,131 @@
-
Go to the documentation of this file.
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
19 #href_anchor"code" href="a00107.html">boxread.h"
-
-
-
-
-
24 #href_anchor"code" href="a00725.html">fileerr.h"
-
-
26 #href_anchor"code" href="a00635.html">helpers.h"
-
-
28 #href_anchor"code" href="a00704.html">strngs.h"
-
-
30 #href_anchor"code" href="a00617.html">unichar.h"
-
-
-
33 static const char * kMultiBlobLabelCode =
"WordStr" ;
+
Go to the documentation of this file.
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
18 #href_anchor"code" href="a00167.html">maxpool.h"
+
+
+
+
+
23 :
Reconfig (name, ni, x_scale, y_scale) {
+
+
+
+
+
+
+
+
+
+
-
-
-
-
38 FILE* box_file =
nullptr ;
-
39 if (!(box_file = fopen(filename.
string (),
"rb" ))) {
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
63 return ReadMemBoxes (target_page, skip_blanks, &box_data[0],
-
64 true , boxes, texts, box_texts,
-
-
-
-
-
69 bool ReadMemBoxes (
int target_page,
bool skip_blanks,
const char * box_data,
-
70 bool continue_on_failure,
-
-
-
-
-
-
-
77 box_str.
split (
'\n' , &lines);
-
78 if (lines.
empty ())
return false ;
-
-
80 for (
int i = 0; i < lines.
size (); ++i) {
-
-
-
-
-
85 if (continue_on_failure)
-
-
-
-
-
90 if (skip_blanks && (utf8_str ==
" " || utf8_str ==
"\t" ))
continue ;
-
91 if (target_page >= 0 && page != target_page)
continue ;
-
92 if (boxes !=
nullptr ) boxes->
push_back (box);
-
93 if (texts !=
nullptr ) texts->
push_back (utf8_str);
-
94 if (box_texts !=
nullptr ) {
-
-
-
-
-
99 if (pages !=
nullptr ) pages->
push_back (page);
-
-
-
102 return num_boxes > 0;
-
-
-
-
-
107 STRING box_filename = image_filename;
-
108 const char *lastdot = strrchr(box_filename.
string (),
'.' );
-
109 if (lastdot !=
nullptr )
-
-
-
112 box_filename +=
".box" ;
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
129 return ReadNextBox (-1, line_number, box_file, utf8_str, bounding_box);
-
-
-
-
-
-
135 bool ReadNextBox (
int target_page,
int *line_number, FILE* box_file,
-
-
-
-
139 char *buffptr = buff;
-
-
141 while (fgets(buff,
sizeof (buff) - 1, box_file)) {
-
-
-
-
145 const auto *ubuf =
reinterpret_cast< const unsigned char *
> (buffptr);
-
146 if (ubuf[0] == 0xef && ubuf[1] == 0xbb && ubuf[2] == 0xbf)
-
-
-
149 if (*buffptr ==
'\n' || *buffptr ==
'\0' )
continue ;
/usr/share/doc/packages/tesseract-ocr-devel/html/a00104.html differs (HTML document, ASCII text)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00104.html 2022-03-29 06:13:00.149587153 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00104.html 2022-03-29 06:13:00.149587153 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/fontinfo.cpp File Reference
+
tesseract: src/lstm/parallel.cpp File Reference
@@ -83,50 +83,22 @@
+
-
#include "fontinfo.h "
-
#href_anchor"el" href="a00752_source.html">bitvector.h"
-
#include "unicity_table.h "
+
#include "parallel.h "
+
#href_anchor"el" href="a00146_source.html">functions.h"
+
#include "networkscratch.h "
Go to the source code of this file.
-
-
-
-
+
&#href_anchor"nav-path" class="navpath">
/usr/share/doc/packages/tesseract-ocr-devel/html/a00104_source.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00104_source.html 2022-03-29 06:13:00.185587290 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00104_source.html 2022-03-29 06:13:00.181587274 +0000
@@ -5,7 +5,7 @@
tesseract: src/ccstruct/fontinfo.cpp Source File
+tesseract: src/lstm/parallel.cpp Source File
Go to the documentation of this file.
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
-
19 #href_anchor"code" href="a00221.html">fontinfo.h"
-
-
21 #href_anchor"code" href="a00701.html">unicity_table.h"
-
-
+
19 #href_anchor"code" href="a00077.html">parallel.h"
+
+
+
+
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
+
+
26 #href_anchor"code" href="a00209.html">networkscratch.h"
+
+
+
+
+
+
+
+
+
+
+
+
+
39 int stack_size =
stack_ .size();
+
40 for (
int i = 1; i < stack_size; ++i) {
+
+
+
+
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
62 uint32_t properties =
get (font_id).properties;
-
63 for (
int f = 0; f < font_set.
size (); ++f) {
-
64 if (
get (font_set[f].fontinfo_id).properties == properties)
-
-
-
-
-
-
-
-
-
73 if (font_set.
empty ())
return false ;
-
74 int first_font = font_set[0].fontinfo_id;
-
75 uint32_t properties =
get (first_font).properties;
-
76 for (
int f = 1; f < font_set.
size (); ++f) {
-
77 if (
get (font_set[f].fontinfo_id).properties != properties)
-
-
-
-
-
-
-
-
-
-
87 for (
int i = 0; i < other->
size (); ++i) {
-
-
89 if (spacing_vec !=
nullptr ) {
-
-
91 if (target_index < 0) {
-
-
-
94 other->
get (i).name =
nullptr ;
-
-
96 delete get (target_index).spacing_vec;
-
97 get (target_index).spacing_vec = other->
get (i).spacing_vec;
-
-
99 other->
get (i).spacing_vec =
nullptr ;
-
-
-
-
-
-
-
-
-
-
109 for (
int i = 0; i <
size (); ++i) {
-
-
-
112 get (i).name =
nullptr ;
-
113 get (i).spacing_vec =
nullptr ;
-
-
-
-
-
-
-
-
-
-
-
124 return strcmp(fi1.
name , fi2.
name ) == 0;
-
-
-
-
-
-
130 for (
int i = 0; i < fs1.
size ; ++i) {
-
-
-
-
-
-
-
-
-
-
-
-
/usr/share/doc/packages/tesseract-ocr-devel/html/a00107.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00107.html 2022-03-29 06:13:00.233587473 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00107.html 2022-03-29 06:13:00.225587443 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/boxread.h File Reference
+
tesseract: src/lstm/generate_lut.py File Reference
@@ -83,567 +83,31 @@
-
#include <cstdio>
-
#include "strngs.h "
-
+
Go to the source code of this file.
-FILE *&#href_anchor"memItemRight" valign="bottom">OpenBoxFile (const STRING &fname) &#href_anchor"memitem:ab3161d7c887ead295045650e34d690fc"> bool ReadAllBoxes (int target_page, bool skip_blanks, const STRING &filename, GenericVector < TBOX > *boxes, GenericVector < STRING > *texts, GenericVector < STRING > *box_texts, GenericVector < int > *pages)&#href_anchor"memitem:a166c16e9d8d3b3eda52ceb38c06dce64"> bool ReadMemBoxes (int target_page, bool skip_blanks, const char *box_data, bool continue_on_failure, GenericVector < TBOX > *boxes, GenericVector < STRING > *texts, GenericVector < STRING > *box_texts, GenericVector < int > *pages)&#href_anchor"memitem:af8465e130ce34e31d743881d41ecba78"> STRING &#href_anchor"memItemRight" valign="bottom">BoxFileName (const STRING &image_filename)&#href_anchor"memitem:a8236e051549f8174e263196602edadc5"> bool ReadNextBox (int *line_number, FILE *box_file, STRING *utf8_str, TBOX *bounding_box)&#href_anchor"memitem:a566c682100d2b5f0a20f39bf99ad2d30"> bool ReadNextBox (int target_page, int *line_number, FILE *box_file, STRING *utf8_str, TBOX *bounding_box)&#href_anchor"memitem:acd464bb54397318c2f6fb8db9e644fd0"> bool ParseBoxFileStr (const char *boxfile_str, int *page_number, STRING *utf8_str, TBOX *bounding_box)&#href_anchor"memitem:a12d56f93e661ff006a382fb8d7314898"> void MakeBoxFileStr (const char *unichar_str, const TBOX &box, int page_num, STRING *box_str)&#href_anchor"memberdecls">
+ namespace &#href_anchor"memItemRight" valign="bottom">generate_lut &#href_anchor"memberdecls">
const int kBoxReadBufSize = 1024 int generate_lut.kTableSize = 4096 float generate_lut.kScaleFactor = 256.0
-
-
-
◆
-
-
-
-
-
- STRING BoxFileName (
- const STRING &&#href_anchor"paramname">image_filename )
-
-
-
-
-
Definition at line 106 of file boxread.cpp .
-
106 {
-
107 STRING box_filename = image_filename;
-
108 const char *lastdot = strrchr(box_filename.
string (),
'.' );
-
109 if (lastdot != nullptr )
-
-
111
-
112 box_filename += ".box" ;
-
113 return box_filename;
-
114 }
-
-
void truncate_at(int32_t index)
-
const char * string() const
-
-
-
-
-
◆
-
-
-
-
-
- void MakeBoxFileStr
- (
- const char *
- unichar_str ,
-
- const TBOX &&#href_anchor"paramname">box ,
-
-
- int
- page_num ,
-
- STRING *&#href_anchor"paramname">box_str
-
- )
-
-
-
-
-
Definition at line 242 of file boxread.cpp .
-
243 {
-
244 *box_str = unichar_str;
-
-
-
-
-
-
250 }
-
-
-
-
-
void add_str_int(const char *str, int number)
-
-
-
-
-
◆
-
-
-
-
-
- FILE * OpenBoxFile
- (
- const STRING &&#href_anchor"paramname">fname )
-
-
-
-
-
Definition at line 36 of file boxread.cpp .
-
36 {
-
-
38 FILE* box_file = nullptr ;
-
39 if (!(box_file = fopen(filename.
string (),
"rb" ))) {
-
-
-
42 }
-
43 return box_file;
-
44 }
-
STRING BoxFileName(const STRING &image_filename)
-
-
constexpr ERRCODE CANTOPENFILE("Can't open file")
-
void error(const char *caller, TessErrorLogCode action, const char *format,...) const
-
-
-
-
-
◆
-
-
-
-
-
- bool ParseBoxFileStr
- (
- const char *
- boxfile_str ,
-
- int *
- page_number ,
-
- STRING *&#href_anchor"paramname">utf8_str ,
-
- TBOX *&#href_anchor"paramname">bounding_box
-
- )
/usr/share/doc/packages/tesseract-ocr-devel/html/a00107.js differs (ASCII text)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00107.js 2020-03-26 00:00:00.000000000 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00107.js 2020-03-26 00:00:00.000000000 +0000
@@ -1,12 +1,5 @@
var a00107 =
[
- [ "BoxFileName", "a00107.html#af8465e130ce34e31d743881d41ecba78", null ],
- [ "MakeBoxFileStr", "a00107.html#a12d56f93e661ff006a382fb8d7314898", null ],
- [ "OpenBoxFile", "a00107.html#a70ccaab4209d69149a2b8a8ff0d6561b", null ],
- [ "ParseBoxFileStr", "a00107.html#acd464bb54397318c2f6fb8db9e644fd0", null ],
- [ "ReadAllBoxes", "a00107.html#ab3161d7c887ead295045650e34d690fc", null ],
- [ "ReadMemBoxes", "a00107.html#a166c16e9d8d3b3eda52ceb38c06dce64", null ],
- [ "ReadNextBox", "a00107.html#a8236e051549f8174e263196602edadc5", null ],
- [ "ReadNextBox", "a00107.html#a566c682100d2b5f0a20f39bf99ad2d30", null ],
- [ "kBoxReadBufSize", "a00107.html#a6a24b858ccf4e9eb9e163b5bcc04b682", null ]
+ [ "kScaleFactor", "a00107.html#aacae173bb776cfcfe58901eafb3284ec", null ],
+ [ "kTableSize", "a00107.html#ae935373c8248b35fb5ead8eeebfa3946", null ]
];
\ No newline at end of file
/usr/share/doc/packages/tesseract-ocr-devel/html/a00107_source.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00107_source.html 2022-03-29 06:13:00.277587642 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00107_source.html 2022-03-29 06:13:00.277587642 +0000
@@ -5,7 +5,7 @@
tesseract: src/ccstruct/boxread.h Source File
+tesseract: src/lstm/generate_lut.py Source File
-
Go to the documentation of this file.
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
20 #href_anchor"line"> 21 #define TESSERACT_CCUTIL_BOXREAD_H_
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
55 bool ReadMemBoxes (
int target_page,
bool skip_blanks,
const char * box_data,
-
56 bool continue_on_failure,
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
78 bool ReadNextBox (
int target_page,
int *line_number, FILE* box_file,
-
-
-
-
-
-
-
-
-
-
-
-
90 #href_anchor"comment">// TESSERACT_CCUTIL_BOXREAD_H_
-
void MakeBoxFileStr(const char *unichar_str, const TBOX &box, int page_num, STRING *box_str)
-
bool ReadMemBoxes(int target_page, bool skip_blanks, const char *box_data, bool continue_on_failure, GenericVector< TBOX > *boxes, GenericVector< STRING > *texts, GenericVector< STRING > *box_texts, GenericVector< int > *pages)
-
const int kBoxReadBufSize
-
FILE * OpenBoxFile(const STRING &fname)
-
bool ReadNextBox(int *line_number, FILE *box_file, STRING *utf8_str, TBOX *bounding_box)
-
bool ReadAllBoxes(int target_page, bool skip_blanks, const STRING &filename, GenericVector< TBOX > *boxes, GenericVector< STRING > *texts, GenericVector< STRING > *box_texts, GenericVector< int > *pages)
-
bool ParseBoxFileStr(const char *boxfile_str, int *page_number, STRING *utf8_str, TBOX *bounding_box)
-
STRING BoxFileName(const STRING &image_filename)
-
-
-
-
+
Go to the documentation of this file.
+
+
+
+
+
+
+
+
+
+
12 print(
"// Generated code with lookup tables" )
+
13 print(
'#include "functions.h"' )
+
14 print(
"namespace tesseract {" )
+
+
16 print(
"const double TanhTable[] = {" )
+
17 for i
in range(kTableSize):
+
18 print(
" %a," % math.tanh(i / kScaleFactor))
+
+
+
21 print(
"const double LogisticTable[] = {" )
+
22 for i
in range(kTableSize):
+
23 print(
" %a," % (1 / (1 + math.exp(-i / kScaleFactor))))
+
+
25 print(
"} // namespace tesseract." )
tesseract: src/ccstruct/ccstruct.h File Reference
+tesseract: src/lstm/fullyconnected.cpp File Reference
-
#include "cutil_class.h "
+
#include "fullyconnected.h "
+
#href_anchor"el" href="a00146_source.html">functions.h"
+
#include "networkscratch.h "
Go to the source code of this file.
-class &#href_anchor"memItemRight" valign="bottom">tesseract::CCStruct &#href_anchor"memberdecls">
namespace tesseract &#href_anchor"nav-path" class="navpath">
+ namespace &#href_anchor"memItemRight" valign="bottom">tesseract &#href_anchor"memberdecls">
+ const int kNumThreads = 1
+
+
+
◆
+
+
+
+
+
+ const int kNumThreads = 1
+
+
+
+
+
+
tesseract: src/ccstruct/ccstruct.h Source File
+tesseract: src/lstm/fullyconnected.cpp Source File
Go to the documentation of this file.
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
-
19 #href_anchor"line"> 20 #define TESSERACT_CCSTRUCT_CCSTRUCT_H_
-
-
-
-
-
-
-
-
+
19 #href_anchor"code" href="a00113.html">fullyconnected.h"
+
+
+
+
+
+
+
+
+
28 #href_anchor"code" href="a00209.html">networkscratch.h"
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
static const double kXHeightCapRatio
-
static const double kXHeightFraction
-
-
static const double kDescenderFraction
-
static const double kAscenderFraction
-
-
+
+
+
+
+
+
+
+
+
+
+
+
41 :
Network (
type , name, ni, no), external_source_(nullptr), int_mode_(false) {
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
90 no_ = code_map.size();
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
122 int width = input.
Width ();
+
+
+
+
+
+
+
+
+
+
+
133 temp_lines[i].Init(
no_ , scratch);
+
134 curr_input[i].Init(
ni_ , scratch);
+
+
/usr/share/doc/packages/tesseract-ocr-devel/html/a00113.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00113.html 2022-03-29 06:13:00.369587994 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00113.html 2022-03-29 06:13:00.365587978 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/publictypes.h File Reference
+
tesseract: src/lstm/fullyconnected.h File Reference
@@ -83,485 +83,26 @@
+
-
+
#include "network.h "
+
#href_anchor"el" href="a00209_source.html">networkscratch.h"
+
Go to the source code of this file.
+class &#href_anchor"memItemRight" valign="bottom">tesseract::FullyConnected &#href_anchor"memberdecls">
namespace &#href_anchor"memItemRight" valign="bottom">tesseract &#href_anchor"memberdecls">
- enum PolyBlockType { PT_UNKNOWN
-, PT_FLOWING_TEXT
-, PT_HEADING_TEXT
-, PT_PULLOUT_TEXT
-, PT_EQUATION
-, PT_INLINE_EQUATION
-, PT_TABLE
-, PT_VERTICAL_TEXT
-, PT_CAPTION_TEXT
-, PT_FLOWING_IMAGE
-, PT_HEADING_IMAGE
-, PT_PULLOUT_IMAGE
-, PT_HORZ_LINE
-, PT_VERT_LINE
-, PT_NOISE
-, PT_COUNT
- enum tesseract::Orientation { tesseract::ORIENTATION_PAGE_UP = 0
-, tesseract::ORIENTATION_PAGE_RIGHT = 1
-, tesseract::ORIENTATION_PAGE_DOWN = 2
-, tesseract::ORIENTATION_PAGE_LEFT = 3
- } enum tesseract::WritingDirection { tesseract::WRITING_DIRECTION_LEFT_TO_RIGHT = 0
-, tesseract::WRITING_DIRECTION_RIGHT_TO_LEFT = 1
-, tesseract::WRITING_DIRECTION_TOP_TO_BOTTOM = 2
- } enum tesseract::TextlineOrder { tesseract::TEXTLINE_ORDER_LEFT_TO_RIGHT = 0
-, tesseract::TEXTLINE_ORDER_RIGHT_TO_LEFT = 1
-, tesseract::TEXTLINE_ORDER_TOP_TO_BOTTOM = 2
- } enum tesseract::PageSegMode { tesseract::PSM_OSD_ONLY = 0
-, tesseract::PSM_AUTO_OSD = 1
-, tesseract::PSM_AUTO_ONLY = 2
-, tesseract::PSM_AUTO = 3
-, tesseract::PSM_SINGLE_COLUMN = 4
-, tesseract::PSM_SINGLE_BLOCK_VERT_TEXT = 5
-, tesseract::PSM_SINGLE_BLOCK = 6
-, tesseract::PSM_SINGLE_LINE = 7
-, tesseract::PSM_SINGLE_WORD = 8
-, tesseract::PSM_CIRCLE_WORD = 9
-, tesseract::PSM_SINGLE_CHAR = 10
-, tesseract::PSM_SPARSE_TEXT = 11
-, tesseract::PSM_SPARSE_TEXT_OSD = 12
-, tesseract::PSM_RAW_LINE = 13
-, tesseract::PSM_COUNT
- enum tesseract::PageIteratorLevel { tesseract::RIL_BLOCK
-, tesseract::RIL_PARA
-, tesseract::RIL_TEXTLINE
-, tesseract::RIL_WORD
-, tesseract::RIL_SYMBOL
- enum tesseract::ParagraphJustification { tesseract::JUSTIFICATION_UNKNOWN
-, tesseract::JUSTIFICATION_LEFT
-, tesseract::JUSTIFICATION_CENTER
-, tesseract::JUSTIFICATION_RIGHT
- } enum tesseract::OcrEngineMode { tesseract::OEM_TESSERACT_ONLY
-, tesseract::OEM_LSTM_ONLY
-, tesseract::OEM_TESSERACT_LSTM_COMBINED
-, tesseract::OEM_DEFAULT
-, tesseract::OEM_COUNT
-
-
-
-
◆
-
-
-
-
Possible types for a POLY_BLOCK or ColPartition. Must be kept in sync with kPBColors in polyblk.cpp and PTIs*Type functions below, as well as kPolyBlockNames in publictypes.cpp . Used extensively by ColPartition, and POLY_BLOCK .
-
-Enumerator
-
-
Definition at line 53 of file publictypes.h .
/usr/share/doc/packages/tesseract-ocr-devel/html/a00113_source.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00113_source.html 2022-03-29 06:13:00.393588085 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00113_source.html 2022-03-29 06:13:00.397588100 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/publictypes.h Source File
+
tesseract: src/lstm/fullyconnected.h Source File
@@ -82,16 +82,16 @@
Go to the documentation of this file.
-
-
+
+
-
+
-
+
@@ -101,246 +101,182 @@
-
-
-
20 #href_anchor"line"> 21 #define TESSERACT_CCSTRUCT_PUBLICTYPES_H_
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
/usr/share/doc/packages/tesseract-ocr-devel/html/a00116.html differs (HTML document, ASCII text)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00116.html 2022-03-29 06:13:00.417588176 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00116.html 2022-03-29 06:13:00.417588176 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/pdblock.h File Reference
+
tesseract: src/lstm/tfnetwork.cpp File Reference
@@ -82,56 +82,17 @@
-
#include "clst.h "
-
#href_anchor"el" href="a00704_source.html">strngs.h"
-
#include "polyblk.h "
-
-
Go to the source code of this file.
-
-class &#href_anchor"memItemRight" valign="bottom">PDBLK &#href_anchor"mdescRight">page block More... class BLOCK_RECT_IT &#href_anchor"memitem:"> class BLOCK_LINE_IT &#href_anchor"mdescRight">rectangle iterator More...
-
-
-
◆
-
-
+
Go to the source code of this file.
/usr/share/doc/packages/tesseract-ocr-devel/html/a00116_source.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00116_source.html 2022-03-29 06:13:00.437588253 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00116_source.html 2022-03-29 06:13:00.437588253 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/pdblock.h Source File
+
tesseract: src/lstm/tfnetwork.cpp Source File
@@ -82,185 +82,176 @@
-
Go to the documentation of this file.
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
19 #href_anchor"line"> 20 #define PDBLOCK_H
+
Go to the documentation of this file.
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
-
-
23 #href_anchor"code" href="a00704.html">strngs.h"
-
+
22 #href_anchor"line"> 23 #include "input.h "
+
24 #href_anchor"code" href="a00209.html">networkscratch.h"
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
50 void set_sides(ICOORDELT_LIST *left, ICOORDELT_LIST *right);
+
26 using tensorflow::Status;
+
27 using tensorflow::Tensor;
+
28 using tensorflow::TensorShape;
+
+
+
+
+
+
34 int TFNetwork::InitFromProtoStr(
const std::string& proto_str) {
+
35 if (!model_proto_.ParseFromString(proto_str))
return 0;
+
36 return InitFromProto();
+
+
+
+
+
+
+
43 std::string proto_str;
+
44 model_proto_.SerializeToString(&proto_str);
+
+
+
47 memcpy(&data[0], proto_str.data(), proto_str.size());
+
+
+
-
-
-
-
-
-
-
61 bottom_left = box.botleft();
-
62 top_right = box.topright();
-
-
-
-
67 int index ()
const {
return index_; }
-
-
-
-
-
74 void move(
const ICOORD vec);
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
121 return left_it.cycled_list() && right_it.cycled_list();
-
-
-
-
-
-
130 bleft =
ICOORD (left_it.data ()->x (), ymin);
-
-
132 tright =
ICOORD (right_it.data ()->x (), ymax);
-
-
-
-
-
-
138 PDBLK * block =
nullptr ;
-
139 ICOORDELT_IT left_it;
-
140 ICOORDELT_IT right_it;
-
+
+
+
+
+
+
57 if (!model_proto_.ParseFromArray(&data[0], data.
size ())) {
+
+
+
60 return InitFromProto();
+
+
+
+
+
65 void TFNetwork::Forward(
bool debug,
const NetworkIO& input,
+
66 const TransposedArray* input_transpose,
+
67 NetworkScratch* scratch, NetworkIO* output) {
+
68 std::vector<std::pair<std::string, Tensor>> tf_inputs;
+
69 int depth = input_shape_.depth();
+
+
+
72 const StrideMap& stride_map = input.stride_map();
+
+
+
/usr/share/doc/packages/tesseract-ocr-devel/html/a00119.html differs (HTML document, ASCII text)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00119.html 2022-03-29 06:13:00.461588345 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00119.html 2022-03-29 06:13:00.465588360 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/ocrrow.h File Reference
+
tesseract: src/lstm/tfnetwork.pb.h File Reference
@@ -83,26 +83,78 @@
+
-
#include <cstdint>
-
#include <cstdio>
-
#include "elst.h "
-
#href_anchor"el" href="a00233_source.html">quspline.h"
-
#include "rect.h "
-
#href_anchor"el" href="a00794_source.html">scrollview.h"
-
#include "werd.h "
+
#include <string>#include <google/protobuf/stubs/common.h>#include <google/protobuf/io/coded_stream.h>#include <google/protobuf/arena.h>#include <google/protobuf/arenastring.h>#include <google/protobuf/generated_message_table_driven.h>#include <google/protobuf/generated_message_util.h>#include <google/protobuf/inlined_string_field.h>#include <google/protobuf/metadata.h>#include <google/protobuf/message.h>#include <google/protobuf/repeated_field.h>#include <google/protobuf/extension_set.h>#include <google/protobuf/unknown_field_set.h>#include "tensorflow/core/framework/graph.pb.h"
Go to the source code of this file.
+
+
+
◆
+
+
+
+
+
+ #define PROTOBUF_INTERNAL_EXPORT_protobuf_tfnetwork_2eproto
+
+
+
+
+
+
+
+
Missing in old package: /usr/share/doc/packages/tesseract-ocr-devel/html/a00119.js
Missing in old package: /usr/share/doc/packages/tesseract-ocr-devel/html/a00119.js
/usr/share/doc/packages/tesseract-ocr-devel/html/a00119_source.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00119_source.html 2022-03-29 06:13:00.493588467 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00119_source.html 2022-03-29 06:13:00.509588528 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/ocrrow.h Source File
+
tesseract: src/lstm/tfnetwork.pb.h Source File
@@ -82,229 +82,847 @@
-
Go to the documentation of this file.
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
20 #href_anchor"line"> 21 #define OCRROW_H
-
-
-
-
-
26 #href_anchor"code" href="a00233.html">quspline.h"
-
-
28 #href_anchor"code" href="a00794.html">scrollview.h"
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
62 return static_cast< float > (baseline.
y (xpos));
-
-
-
-
-
-
68 xheight = new_xheight;
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
112 return has_drop_cap_;
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
+
Go to the documentation of this file.
+
+
+
4 #href_anchor"line"> 5 #define PROTOBUF_INCLUDED_tfnetwork_2eproto
+
+
+
+
9 #include <google/protobuf/stubs/common.h>
+
+
11 #if GOOGLE_PROTOBUF_VERSION < 3006001
+
12 #error This file was generated by a newer version of protoc which is
+
13 #error incompatible with your Protocol Buffer headers. Please update
+
+
+
16 #if 3006001 < GOOGLE_PROTOBUF_MIN_PROTOC_VERSION
+
17 #error This file was generated by an older version of protoc which is
+
18 #error incompatible with your Protocol Buffer headers. Please
+
19 #error regenerate this file with a newer version of protoc.
+
+
+
22 #include <google/protobuf/io/coded_stream.h>
+
23 #include <google/protobuf/arena.h>
+
24 #include <google/protobuf/arenastring.h>
+
25 #include <google/protobuf/generated_message_table_driven.h>
+
26 #include <google/protobuf/generated_message_util.h>
+
27 #include <google/protobuf/inlined_string_field.h>
+
28 #include <google/protobuf/metadata.h>
+
29 #include <google/protobuf/message.h>
+
30 #include <google/protobuf/repeated_field.h>
+
31 #include <google/protobuf/extension_set.h>
+
32 #include <google/protobuf/unknown_field_set.h>
+
33 #include "tensorflow/core/framework/graph.pb.h"
+
+
35 #define PROTOBUF_INTERNAL_EXPORT_protobuf_tfnetwork_2eproto
+
+
+
+
+
40 static const ::google::protobuf::internal::ParseTableField
entries [];
+
41 static const ::google::protobuf::internal::AuxillaryParseTableField
aux [];
+
42 static const ::google::protobuf::internal::ParseTable
schema [1];
+
+
+
45 static const ::google::protobuf::uint32
offsets [];
+
+
+
+
+
+
51 class TFNetworkModelDefaultTypeInternal;
+
+
+
+
/usr/share/doc/packages/tesseract-ocr-devel/html/a00122.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00122.html 2022-03-29 06:13:00.549588681 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00122.html 2022-03-29 06:13:00.533588620 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/polyaprx.cpp File Reference
+
tesseract: src/lstm/convolve.cpp File Reference
@@ -83,883 +83,22 @@
+
-
#include "polyaprx.h "
-
#href_anchor"el" href="a00068_source.html">blobs.h"
-
#include "coutln.h "
-
#href_anchor"el" href="a00647_source.html">errcode.h"
-
#include "mod128.h "
-
#href_anchor"el" href="a00626_source.html">params.h"
-
#include "points.h "
-
#href_anchor"el" href="a00062_source.html">rect.h"
-
#include "tprintf.h "
+
#include "convolve.h "
+
#href_anchor"el" href="a00209_source.html">networkscratch.h"
+
#include "serialis.h "
Go to the source code of this file.
-TESSLINE *&#href_anchor"memItemRight" valign="bottom">ApproximateOutline (bool allow_detailed_fx, C_OUTLINE *c_outline)&#href_anchor"memitem:aff4b13bfd414b463da57ae26acc0c6f3"> EDGEPT *&#href_anchor"memItemRight" valign="bottom">edgesteps_to_edgepts (C_OUTLINE *c_outline, EDGEPT edgepts[])&#href_anchor"memitem:a90ebde0147c2a36a0b1a216645a4eed8"> void fix2 (EDGEPT *start, int area)&#href_anchor"memitem:ab3ec2f619a85e4d8174d3c54084705f7"> EDGEPT *&#href_anchor"memItemRight" valign="bottom">poly2 (EDGEPT *startpt, int area)&#href_anchor"memitem:aa0f03bd75099af8feca58259b9f3f48d"> void cutline (EDGEPT *first, EDGEPT *last , int area)
-
-
-
◆
-
-
-
-
-
- #define approx_dist 15
-
-
-
-
-
-
◆
-
-
-
-
-
- #define DIR 2 /*direction of run */
-
-
-
-
-
-
◆
-
-
-
-
-
- #define FASTEDGELENGTH 256
-
-
-
-
-
-
◆
-
-
-
-
-
- #define FIXED 4 /*OUTLINE point is fixed */
-
-
-
-
-
-
◆
-
-
-
-
-
- #define fixed_dist 20
-
-
-
-
-
-
◆
-
-
-
-
◆
-
-
-
-
-
- #define RUNLENGTH 1 /*length of run */
-
-
-
-
-
-
-
◆
-
-
-
-
-
- TESSLINE * ApproximateOutline (
- bool&#href_anchor"paramname">allow_detailed_fx ,
-
-
- C_OUTLINE *&#href_anchor"paramname">c_outline tesseract: src/ccstruct/polyaprx.cpp Source File
+tesseract: src/lstm/convolve.cpp Source File
-
Go to the documentation of this file.
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
19 #href_anchor"code" href="a00047.html">polyaprx.h"
-
-
-
22 #href_anchor"code" href="a00215.html">coutln.h"
-
-
24 #href_anchor"code" href="a00050.html">mod128.h"
-
-
26 #href_anchor"code" href="a00158.html">points.h"
-
-
28 #href_anchor"code" href="a00668.html">tprintf.h"
-
-
30 #define FASTEDGELENGTH 256
-
-
32 static BOOL_VAR (poly_debug,
false ,
"Debug old poly" );
-
33 static BOOL_VAR (poly_wide_objects_better,
true ,
-
34 "More accurate approx on wide things" );
-
-
-
-
+
Go to the documentation of this file.
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
21 #href_anchor"code" href="a00098.html">convolve.h"
+
+
+
24 #href_anchor"code" href="a00425.html">serialis.h"
+
+
+
+
+
+
30 half_x_(half_x), half_y_(half_y) {
+
+
+
+
+
+
+
+
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
65 EDGEPT * edgepts = stack_edgepts;
-
-
-
-
-
-
-
-
73 if (!poly_wide_objects_better && loop_box.
width () > area)
-
74 area = loop_box.
width ();
-
-
-
-
-
-
-
81 EDGEPT * prev_result =
nullptr ;
-
-
-
-
85 new_pt->prev = prev_result;
-
86 if (prev_result ==
nullptr ) {
-
-
-
89 prev_result->
next = new_pt;
-
90 new_pt->
prev = prev_result;
-
-
92 if (allow_detailed_fx) {
-
-
-
-
-
-
98 edgept = edgept->
next ;
-
-
100 while (edgept != startpt);
-
101 prev_result->
next = result;
-
102 result->
prev = prev_result;
-
103 if (edgepts != stack_edgepts)
-
-
-
-
-
-
-
-
-
-
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
58 int t = dest_index.
t ();
+
+
+
+
+
+
+
+
+
+
+
+
/usr/share/doc/packages/tesseract-ocr-devel/html/a00125.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00125.html 2022-03-29 06:13:00.621588957 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00125.html 2022-03-29 06:13:00.625588972 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/params_training_featdef.h File Reference
+
tesseract: src/lstm/tfnetwork.pb.cc File Reference
@@ -85,79 +85,91 @@
-
+
Macros |
+
Functions |
+
Variables
+
-
#include "genericvector.h "
-
#href_anchor"el" href="a00704_source.html">strngs.h"
-
-
Go to the source code of this file.
+
#include "tfnetwork.pb.h "
+
#href_anchor"a00125_source.html">Go to the source code of this file.
struct &#href_anchor"memItemRight" valign="bottom">tesseract::ParamsTrainingHypothesis &#href_anchor"memitem:"> class tesseract::ParamsTrainingBundle class tesseract::TFNetworkModelDefaultTypeInternal &#href_anchor"memitem:"> struct protobuf_tfnetwork_2eproto::StaticDescriptorInitializer &#href_anchor"memberdecls">
namespace tesseract namespace protobuf_tensorflow_2fcore_2fframework_2fgraph_2eproto &#href_anchor"memitem:a01608"> namespace tesseract &#href_anchor"memitem:a01611"> namespace protobuf_tfnetwork_2eproto &#href_anchor"memitem:a01612"> namespace google &#href_anchor"memitem:a01613"> namespace google::protobuf &#href_anchor"memberdecls">
- using tesseract::ParamsTrainingHypothesisList = GenericVector < ParamsTrainingHypothesis >&#href_anchor"memberdecls">
- enum tesseract::kParamsTrainingFeatureType { tesseract::PTRAIN_DIGITS_SHORT
-, tesseract::PTRAIN_DIGITS_MED
-, tesseract::PTRAIN_DIGITS_LONG
-, tesseract::PTRAIN_NUM_SHORT
-, tesseract::PTRAIN_NUM_MED
-, tesseract::PTRAIN_NUM_LONG
-, tesseract::PTRAIN_DOC_SHORT
-, tesseract::PTRAIN_DOC_MED
-, tesseract::PTRAIN_DOC_LONG
-, tesseract::PTRAIN_DICT_SHORT
-, tesseract::PTRAIN_DICT_MED
-, tesseract::PTRAIN_DICT_LONG
-, tesseract::PTRAIN_FREQ_SHORT
-, tesseract::PTRAIN_FREQ_MED
-, tesseract::PTRAIN_FREQ_LONG
-, tesseract::PTRAIN_SHAPE_COST_PER_CHAR
-, tesseract::PTRAIN_NGRAM_COST_PER_CHAR
-, tesseract::PTRAIN_NUM_BAD_PUNC
-, tesseract::PTRAIN_NUM_BAD_CASE
-, tesseract::PTRAIN_XHEIGHT_CONSISTENCY
-, tesseract::PTRAIN_NUM_BAD_CHAR_TYPE
-, tesseract::PTRAIN_NUM_BAD_SPACING
-, tesseract::PTRAIN_NUM_BAD_FONT
-, tesseract::PTRAIN_RATING_PER_CHAR
-, tesseract::PTRAIN_NUM_FEATURE_TYPES
- #define DO_ (EXPRESSION) if (!GOOGLE_PREDICT_TRUE(EXPRESSION)) goto failure
+
+◆
+
+
+
+ #define DO_
+ (
+
+ EXPRESSION )
+ if (!GOOGLE_PREDICT_TRUE(EXPRESSION)) goto failure
+
+
+
+
+
+
/usr/share/doc/packages/tesseract-ocr-devel/html/a00125.js differs (ASCII text)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00125.js 2020-03-26 00:00:00.000000000 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00125.js 2020-03-26 00:00:00.000000000 +0000
@@ -1,34 +1,19 @@
var a00125 =
[
- [ "tesseract::ParamsTrainingHypothesis", "a02813.html", "a02813" ],
- [ "tesseract::ParamsTrainingBundle", "a02817.html", "a02817" ],
- [ "ParamsTrainingHypothesisList", "a00125.html#a008f78a2d77f64f92e6bd4329409f860", null ],
- [ "kParamsTrainingFeatureType", "a00125.html#a996f57000ba7e945e304ca3a5268b467", [
- [ "PTRAIN_DIGITS_SHORT", "a00125.html#a996f57000ba7e945e304ca3a5268b467a1500cece41ad4b1311e252f2b84d2650", null ],
- [ "PTRAIN_DIGITS_MED", "a00125.html#a996f57000ba7e945e304ca3a5268b467a8e5df70e0a4feb520442eea38ad42581", null ],
- [ "PTRAIN_DIGITS_LONG", "a00125.html#a996f57000ba7e945e304ca3a5268b467ae94102012c081e339674591a1ea39423", null ],
- [ "PTRAIN_NUM_SHORT", "a00125.html#a996f57000ba7e945e304ca3a5268b467a9993582b2b81198845d05ba22a9bdb34", null ],
- [ "PTRAIN_NUM_MED", "a00125.html#a996f57000ba7e945e304ca3a5268b467a70e53977ac4926e96923c104e4da104e", null ],
- [ "PTRAIN_NUM_LONG", "a00125.html#a996f57000ba7e945e304ca3a5268b467a4e3a359d1accb5c88630a86193085b30", null ],
- [ "PTRAIN_DOC_SHORT", "a00125.html#a996f57000ba7e945e304ca3a5268b467abb1f75943b9c111831107bd17c327062", null ],
- [ "PTRAIN_DOC_MED", "a00125.html#a996f57000ba7e945e304ca3a5268b467a073cad7d2dddd23f878513b197f70836", null ],
- [ "PTRAIN_DOC_LONG", "a00125.html#a996f57000ba7e945e304ca3a5268b467a28c22a50dfa668c9d74a39cfab211b10", null ],
- [ "PTRAIN_DICT_SHORT", "a00125.html#a996f57000ba7e945e304ca3a5268b467a4ed49f4fd757ded1bf8eca9aecdbeab7", null ],
- [ "PTRAIN_DICT_MED", "a00125.html#a996f57000ba7e945e304ca3a5268b467aae16de223a8d019d2847a7bdae6c8d5d", null ],
- [ "PTRAIN_DICT_LONG", "a00125.html#a996f57000ba7e945e304ca3a5268b467a70a08810d2d0222f222ce03f5ad0efa8", null ],
- [ "PTRAIN_FREQ_SHORT", "a00125.html#a996f57000ba7e945e304ca3a5268b467a5d90fce75b366d7e15fa7a69d83357c2", null ],
- [ "PTRAIN_FREQ_MED", "a00125.html#a996f57000ba7e945e304ca3a5268b467a695424d03a50adf3fe6a1998acbe3f87", null ],
- [ "PTRAIN_FREQ_LONG", "a00125.html#a996f57000ba7e945e304ca3a5268b467aea70fe3d96122e670ec9e5a4152d93d9", null ],
- [ "PTRAIN_SHAPE_COST_PER_CHAR", "a00125.html#a996f57000ba7e945e304ca3a5268b467a8a4552af7b096cd4e07365b0142b39b2", null ],
- [ "PTRAIN_NGRAM_COST_PER_CHAR", "a00125.html#a996f57000ba7e945e304ca3a5268b467a232084d7e65391f3c7e6b5709a989aa9", null ],
- [ "PTRAIN_NUM_BAD_PUNC", "a00125.html#a996f57000ba7e945e304ca3a5268b467ac5449fec1bf1e8ce13f0c4b534aa9081", null ],
- [ "PTRAIN_NUM_BAD_CASE", "a00125.html#a996f57000ba7e945e304ca3a5268b467a9638fa9da907b679b1ec6ceacc49a625", null ],
- [ "PTRAIN_XHEIGHT_CONSISTENCY", "a00125.html#a996f57000ba7e945e304ca3a5268b467ac1573ea4f6bb4964d253022e801a214e", null ],
- [ "PTRAIN_NUM_BAD_CHAR_TYPE", "a00125.html#a996f57000ba7e945e304ca3a5268b467a2b5b59226e9e2688f6d620a551cc14c9", null ],
- [ "PTRAIN_NUM_BAD_SPACING", "a00125.html#a996f57000ba7e945e304ca3a5268b467abf6ca295b288dca43c10b9022372a1db", null ],
- [ "PTRAIN_NUM_BAD_FONT", "a00125.html#a996f57000ba7e945e304ca3a5268b467adb38a8f09fbe5d6515936780e6cc2561", null ],
- [ "PTRAIN_RATING_PER_CHAR", "a00125.html#a996f57000ba7e945e304ca3a5268b467a0f4d3799e004120bbaeaf1b6194264a8", null ],
- [ "PTRAIN_NUM_FEATURE_TYPES", "a00125.html#a996f57000ba7e945e304ca3a5268b467a1f9a0441244ce25f22a62f3b17cf9d2e", null ]
- ] ],
- [ "ParamsTrainingFeatureByName", "a00125.html#a12cc58bffc8cbb802685ce4cb25aff37", null ]
+ [ "tesseract::TFNetworkModelDefaultTypeInternal", "a03781.html", "a03781" ],
+ [ "protobuf_tfnetwork_2eproto::StaticDescriptorInitializer", "a03785.html", "a03785" ],
+ [ "DO_", "a00125.html#a4c80c2bbfbdfa57f53e1537bbb6567bb", null ],
+ [ "AddDescriptors", "a00125.html#a3ec7cfb9e88f09e708a5a4f047229546", null ],
+ [ "AddDescriptorsImpl", "a00125.html#ace493b5178f96917cc09c67f302931b7", null ],
+ [ "Arena::CreateMaybeMessage< ::tesseract::TFNetworkModel >", "a00125.html#a9df944067993231d53c46e77a45cbb04", null ],
+ [ "GOOGLE_PROTOBUF_ATTRIBUTE_SECTION_VARIABLE", "a00125.html#a406ce7e5fd3d96cf9b1530706b6311b3", null ],
+ [ "InitDefaults", "a00125.html#acc8a807eb411d14ffd9941ec483fe4cd", null ],
+ [ "protobuf_AssignDescriptors", "a00125.html#a1acd47ad74e0d7ecf3606c825c1137db", null ],
+ [ "protobuf_AssignDescriptorsOnce", "a00125.html#aaae0e81baab74a68da48bb959f65acdc", null ],
+ [ "protobuf_RegisterTypes", "a00125.html#a99fc00b72fac5fd7300a6443e1793a28", null ],
+ [ "_TFNetworkModel_default_instance_", "a00125.html#ab05393ea0f1b6367cfad5d6e921b781f", null ],
+ [ "file_level_metadata", "a00125.html#ad49546c08eddfee0e8b8f24362dcb99f", null ],
+ [ "scc_info_GraphDef", "a00125.html#ac93df3bafe4e4cdce47861b836515eae", null ],
+ [ "scc_info_TFNetworkModel", "a00125.html#a730c481c2a2787c6c18ff417af24b40a", null ],
+ [ "static_descriptor_initializer", "a00125.html#aa47e656b54c5435cda1a7b666c0aded1", null ]
];
\ No newline at end of file
/usr/share/doc/packages/tesseract-ocr-devel/html/a00125_source.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00125_source.html 2022-03-29 06:13:00.665589125 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00125_source.html 2022-03-29 06:13:00.685589201 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/params_training_featdef.h Source File
+
tesseract: src/lstm/tfnetwork.pb.cc Source File
@@ -82,212 +82,1031 @@
-
Go to the documentation of this file.
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
20 #href_anchor"line"> 21 #define TESSERACT_WORDREC_PARAMS_TRAINING_FEATDEF_H_
-
-
-
24 #href_anchor"code" href="a00704.html">strngs.h"
-
-
-
-
-
29 static const int kMaxSmallWordUnichars = 3;
-
30 static const int kMaxMediumWordUnichars = 6;
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
73 static const char *
const kParamsTrainingFeatureTypeName[] = {
-
74 "PTRAIN_DIGITS_SHORT" ,
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
89 "PTRAIN_SHAPE_COST_PER_CHAR" ,
-
90 "PTRAIN_NGRAM_COST_PER_CHAR" ,
-
91 "PTRAIN_NUM_BAD_PUNC" ,
-
92 "PTRAIN_NUM_BAD_CASE" ,
-
93 "PTRAIN_XHEIGHT_CONSISTENCY" ,
-
94 "PTRAIN_NUM_BAD_CHAR_TYPE" ,
-
95 "PTRAIN_NUM_BAD_SPACING" ,
-
96 "PTRAIN_NUM_BAD_FONT" ,
-
97 "PTRAIN_RATING_PER_CHAR" ,
-
+
Go to the documentation of this file.
+
+
+
4 #href_anchor"code" href="a00119.html">tfnetwork.pb.h"
+
+
+
+
8 #include <google/protobuf/stubs/common.h>
+
9 #include <google/protobuf/stubs/port.h>
+
10 #include <google/protobuf/io/coded_stream.h>
+
11 #include <google/protobuf/wire_format_lite_inl.h>
+
12 #include <google/protobuf/descriptor.h>
+
13 #include <google/protobuf/generated_message_reflection.h>
+
14 #include <google/protobuf/reflection_ops.h>
+
15 #include <google/protobuf/wire_format.h>
+
+
17 #ifdef GOOGLE_PROTOBUF_ENFORCE_UNIQUENESS
+
18 #include "third_party/protobuf/version.h"
+
+
+
+
+
23 extern PROTOBUF_INTERNAL_EXPORT_protobuf_tensorflow_2fcore_2fframework_2fgraph_2eproto ::google::protobuf::internal::SCCInfo<3>
scc_info_GraphDef ;
+
+
+
+
+
28 ::google::protobuf::internal::ExplicitlyConstructed<TFNetworkModel>
+
+
+
+
+
33 static void InitDefaultsTFNetworkModel() {
+
34 GOOGLE_PROTOBUF_VERIFY_VERSION;
+
+
+
+
+
39 ::google::protobuf::internal::OnShutdownDestroyMessage(ptr);
+
+
+
+
+
+
45 {{ATOMIC_VAR_INIT(::google::protobuf::internal::SCCInfoBase::kUninitialized), 1, InitDefaultsTFNetworkModel}, {
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
78 static ::google::protobuf::Message
const *
const file_default_instances[] = {
+
+
+
+
+
+
+
/usr/share/doc/packages/tesseract-ocr-devel/html/a00128.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00128.html 2022-03-29 06:13:00.709589293 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00128.html 2022-03-29 06:13:00.705589278 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/blobs.cpp File Reference
+
tesseract: src/lstm/stridemap.cpp File Reference
@@ -83,258 +83,19 @@
+
-
#include "blobs.h "
-
#href_anchor"el" href="a00110_source.html">ccstruct.h"
-
#include "clst.h "
-
#href_anchor"el" href="a00635_source.html">helpers.h"
-
#include "linlsq.h "
-
#href_anchor"el" href="a00032_source.html">normalis.h"
-
#include "ocrblock.h "
-
#href_anchor"el" href="a00119_source.html">ocrrow.h"
-
#include "points.h "
-
#href_anchor"el" href="a00047_source.html">polyaprx.h"
-
#include "werd.h "
+
#include "stridemap.h "
#href_anchor"a00128_source.html">Go to the source code of this file.
-
-◆
-
-
-
- void divide_blobs
- (
- TBLOB *&#href_anchor"paramname">blob ,
-
- TBLOB *&#href_anchor"paramname">other_blob ,
-
- bool
- italic_blob ,
-
- const TPOINT &&#href_anchor"paramname">location
-
-
- )
-
-
-
-
-
Definition at line 962 of file blobs.cpp .
-
963 {
-
-
-
-
-
968
-
-
-
971 int location_prod = location.
cross (vertical);
-
972
-
973 while (outline != nullptr ) {
-
-
-
-
977 int mid_prod = mid_pt.cross(vertical);
-
978 if (mid_prod < location_prod) {
-
979
-
980 if (outline1)
-
981 outline1->
next = outline;
-
982 else
-
-
984 outline1 = outline;
-
985 } else {
-
986
-
987 if (outline2)
-
988 outline2->
next = outline;
-
989 else
-
-
991 outline2 = outline;
-
992 }
-
993 outline = outline->
next ;
-
994 }
-
995
-
996 if (outline1) outline1->
next =
nullptr ;
-
997 if (outline2) outline2->
next =
nullptr ;
-
998 }
-
const TPOINT kDivisibleVerticalUpright(0, 1)
-
const TPOINT kDivisibleVerticalItalic(1, 5)
-
-
-
-
int cross(const TPOINT &other) const
-
-
-
-
-
-
-
-
◆
-
-
-
- bool divisible_blob
- (
- TBLOB *&#href_anchor"paramname">blob ,
-
- bool
- italic_blob ,
-
- TPOINT *&#href_anchor"paramname">location
-
- )
-
-
-
-
-
Definition at line 913 of file blobs.cpp .
-
913 {
-
-
915 return false ;
-
916 int max_gap = 0;
-
-
-
-
920 outline1 = outline1->
next ) {
-
921 if (outline1->is_hole) continue ;
-
-
923 static_cast< int16_t> ((outline1->topleft.x + outline1->botright.x) / 2),
-
924 static_cast< int16_t> ((outline1->topleft.y + outline1->botright.y) / 2));
-
925 int mid_prod1 = mid_pt1.cross(vertical);
-
926 int min_prod1, max_prod1;
-
927 outline1->MinMaxCrossProduct(vertical, &min_prod1, &max_prod1);
-
928 for (
TESSLINE * outline2 = outline1->
next ; outline2 !=
nullptr ;
-
929 outline2 = outline2->
next ) {
-
930 if (outline2->is_hole) continue ;
-
931 TPOINT mid_pt2(
static_cast< int16_t
> (
-
932 (outline2->topleft.x + outline2->botright.x) / 2),
-
933 static_cast< int16_t> (
-
934 (outline2->topleft.y + outline2->botright.y) / 2));
-
935 int mid_prod2 = mid_pt2.cross(vertical);
-
936 int min_prod2, max_prod2;
-
937 outline2->MinMaxCrossProduct(vertical, &min_prod2, &max_prod2);
-
938 int mid_gap = abs(mid_prod2 - mid_prod1);
-
939 int overlap =
-
940 std::min(max_prod1, max_prod2) - std::max(min_prod1, min_prod2);
-
941 if (mid_gap - overlap / 4 > max_gap) {
-
942 max_gap = mid_gap - overlap / 4;
-
943 *location = mid_pt1;
-
944 *location += mid_pt2;
-
945 *location /= 2;
-
946 }
/usr/share/doc/packages/tesseract-ocr-devel/html/a00128_source.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00128_source.html 2022-03-29 06:13:00.757589477 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00128_source.html 2022-03-29 06:13:00.737589400 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/blobs.cpp Source File
+
tesseract: src/lstm/stridemap.cpp Source File
@@ -82,1172 +82,206 @@
-
Go to the documentation of this file.
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
+
Go to the documentation of this file.
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
18 #href_anchor"code" href="a00200.html">stridemap.h"
+
-
-
-
-
-
25 #href_anchor"line"> 26 #include "config_auto.h"
-
-
-
-
30 #href_anchor"code" href="a00110.html">ccstruct.h"
-
-
32 #href_anchor"code" href="a00635.html">helpers.h"
-
-
34 #href_anchor"code" href="a00032.html">normalis.h"
-
-
36 #href_anchor"code" href="a00119.html">ocrrow.h"
-
-
38 #href_anchor"code" href="a00047.html">polyaprx.h"
-
+
+
+
+
+
+
26 for (
int index : indices_) {
+
27 if (
index < 0)
return false ;
+
+
+
+
+
+
+
+
+
+
+
38 return MaxIndexOfDim(dimension) == indices_[dimension];
+
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
+
+
+
+
44 int max_index = stride_map_->shape_[dim] - 1;
+
45 if (dim ==
FD_BATCH )
return max_index;
+
+
47 const size_t batch = indices_[
FD_BATCH ];
+
+
49 if (batch >= stride_map_->heights_.size() ||
+
50 stride_map_->heights_[batch] > max_index)
+
+
52 return stride_map_->heights_[batch] - 1;
+
+
54 if (batch >= stride_map_->widths_.size() ||
+
55 stride_map_->widths_[batch] > max_index)
+
+
57 return stride_map_->widths_[batch] - 1;
+
-
-
-
-
-
-
-
-
-
68 TPOINT b0a1, b0a0, a1b1, b0b1, a1a0;
-
-
-
-
-
-
-
-
-
-
-
-
-
81 int b0a1xb0b1 = b0a1.
cross (b0b1);
-
82 int b0b1xb0a0 = b0b1.
cross (b0a0);
-
83 int a1b1xa1a0 = a1b1.
cross (a1a0);
-
-
-
86 int a1a0xa1b0 = -a1a0.
cross (b0a1);
-
-
88 return ((b0a1xb0b1 > 0 && b0b1xb0a0 > 0) ||
-
89 (b0a1xb0b1 < 0 && b0b1xb0a0 < 0)) &&
-
90 ((a1b1xa1a0 > 0 && a1a0xa1b0 > 0) || (a1b1xa1a0 < 0 && a1a0xa1b0 < 0));
-
-
-
-
-
-
96 result->loop = outline;
-
-
-
-
-
-
-
-
-
-
-
-
108 }
while (pt != outline);
-
-
110 result->SetupFromPos();
-
-
-
-
-
-
-
-
-
-
-
121 if (src.
loop !=
nullptr ) {
-
-
-
-
-
126 newpt =
new EDGEPT (*srcpt);
-
127 if (prevpt ==
nullptr ) {
-
/usr/share/doc/packages/tesseract-ocr-devel/html/a00131.html differs (HTML document, ASCII text)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00131.html 2022-03-29 06:13:00.781589568 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00131.html 2022-03-29 06:13:00.781589568 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/matrix.cpp File Reference
+
tesseract: src/lstm/plumbing.cpp File Reference
@@ -82,22 +82,21 @@
-
#include "matrix.h "
-
#href_anchor"el" href="a00611_source.html">callcpp.h"
-
#include "ratngs.h "
-
#href_anchor"el" href="a00668_source.html">tprintf.h"
-
#include "unicharset.h "
+
Go to the source code of this file.
-
-
-
-
+
+namespace &#href_anchor"memItemRight" valign="bottom">tesseract &#href_anchor"nav-path" class="navpath">
/usr/share/doc/packages/tesseract-ocr-devel/html/a00131_source.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00131_source.html 2022-03-29 06:13:00.805589660 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00131_source.html 2022-03-29 06:13:00.813589691 +0000
@@ -5,7 +5,7 @@
tesseract: src/ccstruct/matrix.cpp Source File
+tesseract: src/lstm/plumbing.cpp Source File
-
Go to the documentation of this file.
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
23 #href_anchor"code" href="a00098.html">matrix.h"
-
-
-
26 #href_anchor"code" href="a00155.html">ratngs.h"
-
-
28 #href_anchor"code" href="a00737.html">unicharset.h"
-
-
-
-
-
-
-
-
-
-
38 BLOB_CHOICE_IT b_it(
get (col, row));
-
39 for (b_it.mark_cycle_pt(); !b_it.cycled_list(); b_it.forward()) {
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
62 for (
int col = ind; col >= 0 && col > ind - band_width; --col) {
-
63 if (
array_ [col * band_width + band_width - 1] !=
empty_ ) {
-
-
-
-
-
68 auto * result =
new MATRIX (dim + 1, band_width);
-
-
70 for (
int col = 0; col < dim; ++col) {
-
71 for (
int row = col; row < dim && row < col +
bandwidth (); ++row) {
-
-
-
74 BLOB_CHOICE_LIST* choices =
get (col, row);
-
75 if (choices !=
nullptr ) {
-
-
77 BLOB_CHOICE_IT bc_it(choices);
-
78 for (bc_it.mark_cycle_pt(); !bc_it.cycled_list(); bc_it.forward()) {
-
-
-
-
-
83 result->put(coord.
col , coord.
row , choices);
-
-
-
-
-
-
-
-
-
-
-
-
-
-
97 auto * result =
new MATRIX (dim, band_width);
-
98 for (
int col = 0; col < dim; ++col) {
-
99 for (
int row = col; row < dim && row < col + band_width; ++row) {
-
100 BLOB_CHOICE_LIST* choices =
get (col, row);
-
101 if (choices !=
nullptr ) {
-
102 auto * copy_choices =
new BLOB_CHOICE_LIST;
-
-
104 result->put(col, row, copy_choices);
-
+
Go to the documentation of this file.
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
20 #href_anchor"code" href="a00188.html">plumbing.h"
+
+
+
+
+
+
+
+
+
+
+
+
+
33 for (
int i = 0; i <
stack_ .size(); ++i)
+
+
+
+
+
+
+
+
41 for (
int i = 0; i <
stack_ .size(); ++i)
+
+
+
+
+
+
+
+
+
+
+
52 for (
int i = 0; i <
stack_ .size(); ++i)
+
+
+
+
+
+
+
+
+
61 for (
int i = 0; i <
stack_ .size(); ++i) {
+
+
+
+
+
+
+
+
69 for (
int i = 0; i <
stack_ .size(); ++i)
+
+
+
+
+
+
+
+
77 for (
int i = 0; i <
stack_ .size(); ++i)
/usr/share/doc/packages/tesseract-ocr-devel/html/a00134.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00134.html 2022-03-29 06:13:00.833589767 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00134.html 2022-03-29 06:13:00.833589767 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/pdblock.cpp File Reference
+
tesseract: src/lstm/ctc.h File Reference
@@ -83,102 +83,28 @@
+
-
#include "allheaders.h"
-
#include "pdblock.h "
-
#href_anchor"a00134_source.html">Go to the source code of this file.
-
-constexpr ERRCODE &#href_anchor"memItemRight" valign="bottom">BADBLOCKLINE ("Y coordinate in block out of bounds") constexpr ERRCODE &#href_anchor"memItemRight" valign="bottom">LOSTBLOCKLINE ("Can't find rectangle for line")
-
-◆
-
-
-
- #define BLOCK_LABEL_HEIGHT 150
-
-
-
-
-
Definition at line 30 of file pdblock.cpp .
-
+
#include "genericvector.h "
+
#href_anchor"el" href="a00083_source.html">network.h"
+
#include "networkio.h "
+
#href_anchor"el" href="a00854_source.html">scrollview.h"
-
-
-
-
◆
-
-
-
-
-
-
-
-
- constexpr ERRCODE BADBLOCKLINE
- (
- "Y coordinate in block out of bounds"
- )
-
-
-
-
-constexpr
-
-
-
-
-
-
-
-
◆
-
-
-
-
-
-
-
-
- constexpr ERRCODE LOSTBLOCKLINE
- (
- "Can't find rectangle for line"&#href_anchor"paramname"> )
-
-
-
-
-constexpr
-
-
-
-
-
-
-
-
-
+
Go to the source code of this file.
+
+class &#href_anchor"memItemRight" valign="bottom">tesseract::CTC &#href_anchor"memberdecls">
+ namespace tesseract &#href_anchor"nav-path" class="navpath">
/usr/share/doc/packages/tesseract-ocr-devel/html/a00134_source.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00134_source.html 2022-03-29 06:13:00.861589874 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00134_source.html 2022-03-29 06:13:00.861589874 +0000
@@ -5,7 +5,7 @@
tesseract: src/ccstruct/pdblock.cpp Source File
+tesseract: src/lstm/ctc.h Source File
-
Go to the documentation of this file.
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
21 #href_anchor"line"> 22 #include <memory>
-
-
-
-
-
27 #include "config_auto.h"
-
-
-
30 #define BLOCK_LABEL_HEIGHT 150
-
-
-
-
-
-
-
-
-
-
-
-
-
43 int16_t ymin, int16_t xmax,
-
44 int16_t ymax): box (
ICOORD (xmin, ymin),
ICOORD (xmax, ymax)) {
-
-
46 ICOORDELT_IT left_it = &leftside;
-
47 ICOORDELT_IT right_it = &rightside;
-
-
-
50 left_it.set_to_list (&leftside);
-
51 right_it.set_to_list (&rightside);
-
-
53 left_it.add_to_end (
new ICOORDELT (xmin, ymin));
-
54 left_it.add_to_end (
new ICOORDELT (xmin, ymax));
-
55 right_it.add_to_end (
new ICOORDELT (xmax, ymin));
-
56 right_it.add_to_end (
new ICOORDELT (xmax, ymax));
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
76 left_it.move_to_first();
-
77 left_it.add_list_before(left);
-
-
79 right_it.move_to_first();
-
80 right_it.add_list_before(right);
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
99 if (pt.
x () >= bleft.
x () && pt.
x () <= tright.
x ()
-
100 && pt.
y () >= bleft.
y () && pt.
y () <= tright.
y ())
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
118 for (it.mark_cycle_pt (); !it.cycled_list (); it.forward ())
-
119 *(it.data ()) += vec;
-
-
-
-
123 for (it.mark_cycle_pt (); !it.cycled_list (); it.forward ())
-
124 *(it.data ()) += vec;
-
-
-
-
-
-
-
-
-
133 rotated_box.
rotate (rerotation);
-
134 Pix* pix = pixCreate(rotated_box.
width (), rotated_box.
height (), 1);
-
-
-
-
138 ICOORDELT_LIST polygon;
-
-
-
141 image_block.
rotate (rerotation);
-
-
-
-
-
146 const std::unique_ptr< ICOORDELT_LIST> segments(
-
-
148 if (!segments->empty()) {
-
149 ICOORDELT_IT s_it(segments.get());
-
-
-
152 for (s_it.mark_cycle_pt(); !s_it.cycled_list(); s_it.forward()) {
-
153 int start = s_it.data()->x();
-
154 int xext = s_it.data()->y();
-
-
156 pixRasterop(pix, start - rotated_box.
left (),
-
-
158 xext, 1, PIX_SET,
nullptr , 0, 0);
-
-
-
-
-
-
-
165 pixRasterop(pix, 0, 0, rotated_box.
width (), rotated_box.
height (),
-
166 PIX_SET,
nullptr , 0, 0);
-
-
168 if (mask_box !=
nullptr ) *mask_box = rotated_box;
-
-
-
-
-
-
-
-
-
-
-
179 #ifndef GRAPHICS_DISABLED
-
-
-
-
/usr/share/doc/packages/tesseract-ocr-devel/html/a00137.html differs (HTML document, ASCII text)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00137.html 2022-03-29 06:13:00.885589966 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00137.html 2022-03-29 06:13:00.885589966 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/debugpixa.h File Reference
+
tesseract: src/lstm/input.cpp File Reference
@@ -83,25 +83,33 @@
+
-
#include "allheaders.h"
+
#include "input.h "
+
#href_anchor"el" href="a00665_source.html">imagedata.h"
+
#include "pageres.h "
+
#href_anchor"el" href="a00854_source.html">scrollview.h"
Go to the source code of this file.
+
+
+
+
Missing in old package: /usr/share/doc/packages/tesseract-ocr-devel/html/a00137.js
Missing in old package: /usr/share/doc/packages/tesseract-ocr-devel/html/a00137.js
/usr/share/doc/packages/tesseract-ocr-devel/html/a00137_source.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00137_source.html 2022-03-29 06:13:00.909590058 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00137_source.html 2022-03-29 06:13:00.917590089 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/debugpixa.h Source File
+
tesseract: src/lstm/input.cpp Source File
@@ -82,72 +82,195 @@
-
Go to the documentation of this file. 1 #href_anchor"line"> 2 #define TESSERACT_CCSTRUCT_DEBUGPIXA_H_
-
-
-
-
-
-
-
-
-
-
-
-
14 pixa_ = pixaCreate(0);
-
15 fonts_ = bmfCreate(
nullptr , 14);
-
-
-
-
-
-
-
-
-
-
-
26 void AddPix (
const Pix* pix,
const char * caption) {
-
27 int depth = pixGetDepth(
const_cast< Pix*
> (pix));
-
28 int color = depth < 8 ? 1 : (depth > 8 ? 0x00ff0000 : 0x80);
-
29 Pix* pix_debug = pixAddSingleTextblock(
-
30 const_cast< Pix*
> (pix), fonts_, caption, color, L_ADD_BELOW,
nullptr );
-
31 pixaAddPix(pixa_, pix_debug, L_INSERT);
-
-
-
-
-
-
37 if (pixaGetCount(pixa_) > 0) {
-
38 pixaConvertToPdf(pixa_, 300, 1.0f, 0, 0,
"AllDebugImages" , filename);
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
void WritePDF(const char *filename)
-
void AddPix(const Pix *pix, const char *caption)
-
-
+
Go to the documentation of this file.
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
18 #href_anchor"code" href="a00191.html">input.h"
+
+
20 #include "allheaders.h"
+
+
22 #href_anchor"code" href="a00815.html">pageres.h"
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
59 cached_x_scale_ = factor;
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
75 tprintf (
"Input::Backward should not be called!!\n" );
+
+
+
+
+
+
+
+
+
84 const Network * network,
int min_width,
+
85 TRand * randomizer,
float * image_scale) {
+
+
+
+
+
90 &width, &height,
nullptr );
+
+
92 tprintf (
"Bad pix from ImageData!\n" );
+
+
+
95 if (width < min_width || height < min_width) {
+
96 tprintf (
"Image too small to scale!! (%dx%d vs min width of %d)\n" , width,
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
113 bool color = shape.
depth () == 3;
+
114 Pix* var_pix =
const_cast< Pix*
> (pix);
+
115 int depth = pixGetDepth(var_pix);
+
116 Pix* normed_pix =
nullptr ;
+
+
+
+
+
+
122 normed_pix = pixClone(var_pix);
+
+
124 normed_pix = pixConvertTo32(var_pix);
+
/usr/share/doc/packages/tesseract-ocr-devel/html/a00140.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00140.html 2022-03-29 06:13:00.941590180 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00140.html 2022-03-29 06:13:00.945590195 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/pageres.cpp File Reference
+
tesseract: src/lstm/recodebeam.cpp File Reference
@@ -83,132 +83,29 @@
+
-
#include "pageres.h "
-
#href_anchor"el" href="a00197_source.html">blamer.h"
-
#include "blobs.h "
-
#href_anchor"el" href="a00017_source.html">boxword.h"
-
#include "errcode.h "
-
#href_anchor"el" href="a00179_source.html">ocrblock.h"
-
#include "ocrrow.h "
-
#href_anchor"el" href="a00116_source.html">pdblock.h"
-
#include "polyblk.h "
-
#href_anchor"el" href="a00113_source.html">publictypes.h"
-
#include "seam.h "
-
#href_anchor"el" href="a00089_source.html">stepblob.h"
-
#include "tesscallback.h "
-
#href_anchor"el" href="a00668_source.html">tprintf.h"
+
#include "recodebeam.h "
+
#href_anchor"el" href="a00086_source.html">networkio.h"
+
#include "pageres.h "
+
#href_anchor"el" href="a00479_source.html">unicharcompress.h"
+
#include <deque>
+
#include <map>
+
#include <set>
+
#include <tuple>
+
#include <vector>
+
#include <algorithm>
Go to the source code of this file.
-
-
-
◆
-
-
-
-
-
- CLISTIZE
- (
- BLOCK_RES &#href_anchor"paramname">)
- const
-
-
-
-
-
-
-
-
-
◆
-
-
-
-
-
- const double kMaxLineSizeRatio = 1.25
-
-
-
-
-
-
◆
-
-
-
-
-
- const double kMaxWordGapRatio = 2.0
-
-
-
-
-
-
◆
-
-
-
-
-
- const double kMaxWordSizeRatio = 1.25
-
-
-
-
-
-
◆
-
-
-
-
-
- const int kWordrecMaxNumJoinChunks = 4
-
-
-
-
-
-
-
-
+
+
namespace &#href_anchor"memItemRight" valign="bottom">tesseract
+
&#href_anchor"nav-path" class="navpath">
/usr/share/doc/packages/tesseract-ocr-devel/html/a00140_source.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00140_source.html 2022-03-29 06:13:01.017590471 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00140_source.html 2022-03-29 06:13:00.997590394 +0000
@@ -5,7 +5,7 @@
tesseract: src/ccstruct/pageres.cpp Source File
+tesseract: src/lstm/recodebeam.cpp Source File
-
Go to the documentation of this file.
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
24 #href_anchor"code" href="a00227.html">pageres.h"
-
-
-
-
-
29 #href_anchor"code" href="a00068.html">blobs.h"
-
-
31 #href_anchor"code" href="a00647.html">errcode.h"
-
-
33 #href_anchor"code" href="a00119.html">ocrrow.h"
-
-
35 #href_anchor"code" href="a00056.html">polyblk.h"
-
-
37 #href_anchor"code" href="a00083.html">seam.h"
-
-
39 #href_anchor"code" href="a00770.html">tesscallback.h"
-
-
-
-
-
-
-
-
-
48 static const
double kStopperAmbiguityThresholdGain = 8.0;
-
-
-
51 static const
double kStopperAmbiguityThresholdOffset = 1.5;
-
-
-
-
-
-
-
-
-
-
-
-
-
-
65 static
double StopperAmbigThreshold(
double f1,
double f2) {
-
66 return (f2 - f1) * kStopperAmbiguityThresholdGain -
-
67 kStopperAmbiguityThresholdOffset;
-
-
-
-
-
-
-
-
-
76 bool merge_similar_words,
-
77 BLOCK_LIST *the_block_list,
-
-
-
80 BLOCK_IT block_it(the_block_list);
-
-
82 for (block_it.mark_cycle_pt();
-
83 !block_it.cycled_list(); block_it.forward()) {
-
84 block_res_it.add_to_end(
new BLOCK_RES (merge_similar_words,
-
-
-
-
-
-
-
-
-
-
-
-
-
97 ROW_IT row_it (the_block->
row_list ());
-
-
-
-
-
-
-
-
-
-
-
-
109 for (row_it.mark_cycle_pt(); !row_it.cycled_list(); row_it.forward()) {
-
110 row_res_it.add_to_end(
new ROW_RES (merge_similar_words, row_it.data()));
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
131 bool add_next_word =
false ;
-
-
-
-
135 for (word_it.mark_cycle_pt(); !word_it.cycled_list(); word_it.forward()) {
-
136 auto * word_res =
new WERD_RES (word_it.data());
-
137 word_res->x_height = the_row->
x_height ();
-
-
-
-
141 word_res->part_of_combo =
true ;
-
-
143 }
else if (merge_similar_words) {
-
144 union_box = word_res->word->bounding_box();
-
145 add_next_word = !word_res->word->flag(
W_REP_CHAR ) &&
-
-
147 word_res->odd_size = !add_next_word;
-
-
149 WERD * next_word = word_it.data_relative(1);
-
150 if (merge_similar_words) {
-
-
-
-
-
-
-
-
158 int prev_right = union_box.
right ();
-
159 union_box += next_box;
-
-
-
-
163 add_next_word =
false ;
-
-
-
-
-
-
-
-
171 if (combo ==
nullptr ) {
-
172 copy_word =
new WERD ;
-
173 *copy_word = *(word_it.data());
-
-
-
-
177 word_res_it.add_to_end(combo);
-
-
179 word_res->part_of_combo =
true ;
-
-
/usr/share/doc/packages/tesseract-ocr-devel/html/a00143.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00143.html 2022-03-29 06:13:01.049590593 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00143.html 2022-03-29 06:13:01.045590578 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/blread.h File Reference
+
tesseract: src/lstm/ctc.cpp File Reference
@@ -83,141 +83,25 @@
+
-
#include <cstdint>
-
#include "strngs.h "
+
#include "ctc.h "
+
#href_anchor"el" href="a00392_source.html">genericvector.h"
+
#include "matrix.h "
+
#href_anchor"el" href="a00086_source.html">networkio.h"
+
#include "network.h "
+
#href_anchor"el" href="a00854_source.html">scrollview.h"
Go to the source code of this file.
-bool&#href_anchor"memItemRight" valign="bottom">read_unlv_file (STRING name, int32_t xsize, int32_t ysize, BLOCK_LIST *blocks) &#href_anchor"memitem:a6d4d883ff45201c5b6166b5941cb96a2"> void FullPageBlock (int width, int height, BLOCK_LIST *blocks)
-
-
-
◆
-
-
-
-
-
- void FullPageBlock
- (
- int
- width ,
-
- int
- height ,
-
- BLOCK_LIST *
- blocks
-
- )
-
-
-
-
-
Definition at line 64 of file blread.cpp .
-
64 {
-
65 BLOCK_IT block_it(blocks);
-
66 auto * block =
new BLOCK (
"" ,
true , 0, 0, 0, 0, width, height);
-
67 block_it.add_to_end(block);
-
68 }
-
-
-
-
-
-
◆
-
-
-
-
-
- bool read_unlv_file
- (
- STRING &#href_anchor"paramname">name ,
-
- int32_t
- xsize ,
-
- int32_t
- ysize ,
-
- BLOCK_LIST *
- blocks
-
- )
-
-
-
-
-
Definition at line 32 of file blread.cpp .
-
37 {
-
38 FILE *pdfp;
-
-
40 int x;
-
41 int y;
-
42 int width;
-
43 int height;
-
44 BLOCK_IT block_it = blocks;
-
45
-
-
47 if ((pdfp = fopen (name.
string (),
"rb" )) ==
nullptr ) {
-
48 return false ;
-
49 } else {
-
50 while (
tfscanf (pdfp,
"%d %d %d %d %*s" , &x, &y, &width, &height) >= 4) {
-
51
-
-
53 static_cast< int16_t> (x), static_cast< int16_t> (ysize - y - height),
-
54 static_cast< int16_t> (x + width), static_cast< int16_t> (ysize - y));
-
55
-
56 block_it.add_to_end (block);
-
57 }
-
58 fclose(pdfp);
-
59 }
-
-
61 return true ;
-
62 }
-
-
DLLSYM void tprintf(const char *format,...)
-
int tfscanf(FILE *stream, const char *format,...)
-
const char * string() const
-
-
-
-
-
-
-
+
+
namespace &#href_anchor"memItemRight" valign="bottom">tesseract
+
&#href_anchor"nav-path" class="navpath">
/usr/share/doc/packages/tesseract-ocr-devel/html/a00143_source.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00143_source.html 2022-03-29 06:13:01.069590669 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00143_source.html 2022-03-29 06:13:01.081590715 +0000
@@ -5,7 +5,7 @@
tesseract: src/ccstruct/blread.h Source File
+tesseract: src/lstm/ctc.cpp Source File
-
Go to the documentation of this file.
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
+
Go to the documentation of this file.
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
18 #href_anchor"code" href="a00134.html">ctc.h"
-
20 #href_anchor"line"> 21 #define BLREAD_H
-
-
-
-
-
+
+
+
+
+
+
25 #href_anchor"code" href="a00725.html">matrix.h"
+
-
-
-
-
-
-
-
-
-
-
void FullPageBlock(int width, int height, BLOCK_LIST *blocks)
-
bool read_unlv_file(STRING name, int32_t xsize, int32_t ysize, BLOCK_LIST *blocks)
-
-
+
28 #href_anchor"code" href="a00083.html">network.h"
+
+
+
+
+
+
+
35 const float CTC::kMinProb_ = 1e-12;
+
+
37 const double CTC::kMaxExpArg_ = 80.0;
+
+
39 const double CTC::kMinTotalTimeProb_ = 1e-8;
+
+
41 const double CTC::kMinTotalFinalProb_ = 1e-6;
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
57 std::unique_ptr<CTC> ctc(
new CTC (labels, null_char, outputs));
+
58 if (!ctc->ComputeLabelLimits()) {
+
+
+
+
+
+
64 ctc->ComputeSimpleTargets(&simple_targets);
+
+
66 float bias_fraction = ctc->CalculateBiasFraction();
+
67 simple_targets *= bias_fraction;
+
68 ctc->outputs_ += simple_targets;
+
+
+
+
+
73 ctc->Forward(&log_alphas);
+
74 ctc->Backward(&log_betas);
+
+
76 log_alphas += log_betas;
+
77 ctc->NormalizeSequence(&log_alphas);
+
78 ctc->LabelsToClasses(log_alphas, targets);
+
+
+
+
+
+
+
85 : labels_(labels), outputs_(outputs), null_char_(null_char) {
+
86 num_timesteps_ = outputs.
dim1 ();
+
87 num_classes_ = outputs.
dim2 ();
+
88 num_labels_ = labels_.
size ();
+
+
+
+
+
93 bool CTC::ComputeLabelLimits() {
+
+
+
96 int min_u = num_labels_ - 1;
+
97 if (labels_[min_u] == null_char_) --min_u;
+
98 for (
int t = num_timesteps_ - 1; t >= 0; --t) {
+
99 min_labels_[t] = min_u;
+
+
+
102 if (labels_[min_u] == null_char_ && min_u > 0 &&
+
103 labels_[min_u + 1] != labels_[min_u - 1]) {
+
+
+
+
+
108 int max_u = labels_[0] == null_char_;
+
109 for (
int t = 0; t < num_timesteps_; ++t) {
+
110 max_labels_[t] = max_u;
+
111 if (max_labels_[t] < min_labels_[t])
return false ;
+
112 if (max_u + 1 < num_labels_) {
+
+
114 if (labels_[max_u] == null_char_ && max_u + 1 < num_labels_ &&
+
115 labels_[max_u + 1] != labels_[max_u - 1]) {
+
+
+
+
+
+
+
+
+
+
+
+
127 targets->
Resize (num_timesteps_, num_classes_, 0.0f);
+
+
+
130 ComputeWidthsAndMeans(&half_widths, &means);
+
131 for (
int l = 0; l < num_labels_; ++l) {
+
132 int label = labels_[l];
+
133 float left_half_width = half_widths[l];
+
134 float right_half_width = left_half_width;
+
+
136 if (label == null_char_) {
+
137 if (!NeededNull(l)) {
+
138 if ((l > 0 && mean == means[l - 1]) ||
+
139 (l + 1 < num_labels_ && mean == means[l + 1])) {
+
+
+
+
+
+
145 if (l > 0) left_half_width = mean - means[l - 1];
/usr/share/doc/packages/tesseract-ocr-devel/html/a00146.html differs (HTML document, ASCII text)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00146.html 2022-03-29 06:13:01.101590792 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00146.html 2022-03-29 06:13:01.105590808 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/ocrrow.cpp File Reference
+
tesseract: src/lstm/functions.h File Reference
@@ -82,19 +82,109 @@
-
#include "ocrrow.h "
-
#href_anchor"el" href="a00182_source.html">blobbox.h"
+
Go to the source code of this file.
+
+double tesseract::Tanh (double x) double tesseract::Logistic (double x) template<class Func > void tesseract::FuncInplace (int n, double *inout) template<class Func > void tesseract::FuncMultiply (const double *u, const double *v, int n, double *out) template<typename T > void tesseract::SoftmaxInPlace (int n, T *inout) void tesseract::CopyVector (int n, const double *src, double *dest) void tesseract::AccumulateVector (int n, const double *src, double *dest) void tesseract::MultiplyVectorsInPlace (int n, const double *src, double *inout) void tesseract::MultiplyAccumulate (int n, const double *u, const double *v, double *out) void tesseract::SumVectors (int n, const double *v1, const double *v2, const double *v3, const double *v4, const double *v5, double *sum) template<typename T > void tesseract::ZeroVector (int n, T *vec) template<typename T > void tesseract::ClipVector (int n, T lower, T upper, T *vec) void tesseract::CodeInBinary (int n, int nf, double *vec)
+
+
+
◆
+
+
+
+
+
+ #define DEBUG_DETAIL 0
+
+
+
+
Missing in old package: /usr/share/doc/packages/tesseract-ocr-devel/html/a00146.js
Missing in old package: /usr/share/doc/packages/tesseract-ocr-devel/html/a00146.js
/usr/share/doc/packages/tesseract-ocr-devel/html/a00146_source.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00146_source.html 2022-03-29 06:13:01.137590930 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00146_source.html 2022-03-29 06:13:01.145590961 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/ocrrow.cpp Source File
+
tesseract: src/lstm/functions.h Source File
@@ -82,293 +82,294 @@
-
Go to the documentation of this file.
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
19 #href_anchor"code" href="a00119.html">ocrrow.h"
-
-
-
-
23 #href_anchor"line"> 24 #include "config_auto.h"
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
44 :
baseline (spline_size, xstarts, coeffs),
-
-
-
-
-
-
-
51 descdrop = descenders;
-
52 has_drop_cap_ =
false ;
-
-
-
-
-
-
-
-
-
-
-
+
Go to the documentation of this file.
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
18 #href_anchor"line"> 19 #define TESSERACT_LSTM_FUNCTIONS_H_
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
43 inline double Tanh (
double x) {
+
44 if (x < 0.0)
return -
Tanh (-x);
+
+
46 unsigned index =
static_cast< unsigned > (x);
+
+
+
+
+
51 return tanh_i0 + (tanh_i1 - tanh_i0) * (x - index);
+
+
+
+
55 if (x < 0.0)
return 1.0 -
Logistic (-x);
+
+
57 unsigned index =
static_cast< unsigned > (x);
+
+
+
+
+
62 return l0 + (l1 - l0) * (x - index);
+
-
-
-
-
-
-
-
-
-
-
-
-
-
77 has_drop_cap_ =
false ;
-
-
-
-
-
-
-
-
-
-
87 WERD_IT it(
const_cast< WERD_LIST *
> (&words));
-
88 for (it.mark_cycle_pt(); !it.cycled_list(); it.forward()) {
-
89 box += it.data()->restricted_bounding_box(upper_dots, lower_dots);
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
110 while (!it.at_first ()) {
-
-
-
113 if (left < prev_left) {
-
-
-
-
-
-
-
-
/usr/share/doc/packages/tesseract-ocr-devel/html/a00149.html differs (HTML document, ASCII text)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00149.html 2022-03-29 06:13:01.161591021 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00149.html 2022-03-29 06:13:01.165591037 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/quadlsq.h File Reference
+
tesseract: src/lstm/series.cpp File Reference
@@ -83,20 +83,24 @@
+
-
#include "points.h "
+
#include "series.h "
+
#href_anchor"el" href="a00113_source.html">fullyconnected.h"
+
#include "networkscratch.h "
+
#href_anchor"el" href="a00854_source.html">scrollview.h"
+
#include "tprintf.h "
Go to the source code of this file.
-class &#href_anchor"memItemRight" valign="bottom">QLSQ namespace &#href_anchor"memItemRight" valign="bottom">tesseract &#href_anchor"nav-path" class="navpath">
/usr/share/doc/packages/tesseract-ocr-devel/html/a00149_source.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00149_source.html 2022-03-29 06:13:01.181591098 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00149_source.html 2022-03-29 06:13:01.189591129 +0000
@@ -5,7 +5,7 @@
tesseract: src/ccstruct/quadlsq.h Source File
+tesseract: src/lstm/series.cpp Source File
-
Go to the documentation of this file.
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
20 #href_anchor"line"> 21 #define QUADLSQ_H
-
-
-
-
-
-
-
-
-
-
+
Go to the documentation of this file.
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
19 #href_anchor"code" href="a00179.html">series.h"
+
+
+
22 #href_anchor"code" href="a00209.html">networkscratch.h"
+
+
24 #href_anchor"code" href="a00398.html">tprintf.h"
+
+
+
+
+
+
+
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
void remove(double x, double y)
-
-
-
-
-
-
void add(double x, double y)
-
+
+
+
+
+
37 int stack_size =
stack_ .size();
+
38 for (
int i = 0; i < stack_size; ++i) {
+
39 result =
stack_ [i]->OutputShape(result);
+
+
+
+
+
+
+
+
+
+
49 tprintf (
"Num outputs,weights in Series:\n" );
+
50 for (
int i = 0; i <
stack_ .size(); ++i) {
+
51 int weights =
stack_ [i]->InitWeights(range, randomizer);
+
+
+
+
+
+
+
+
+
+
+
+
+
64 tprintf (
"Num (Extended) outputs,weights in Series:\n" );
+
65 for (
int i = 0; i <
stack_ .size(); ++i) {
+
66 int weights =
stack_ [i]->RemapOutputs(old_no, code_map);
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
81 for (
int i = 0; i <
stack_ .size(); ++i)
+
+
83 return needs_backprop;
+
+
+
+
+
+
+
+
+
+
+
94 for (
int i = 0; i <
stack_ .size(); ++i)
+
+
+
+
+
+
+
+
102 stack_ [0]->CacheXScaleFactor(factor);
+
+
+
+
/usr/share/doc/packages/tesseract-ocr-devel/html/a00152.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00152.html 2022-03-29 06:13:01.213591220 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00152.html 2022-03-29 06:13:01.209591205 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/rejctmap.h File Reference
+
tesseract: src/lstm/functions.cpp File Reference
@@ -83,259 +83,32 @@
+
-
#include <memory>
-
#include "bits16.h "
-
#href_anchor"el" href="a00647_source.html">errcode.h"
-
#include "params.h "
+
Go to the source code of this file.
-enum REJ_FLAGS { R_TESS_FAILURE
-, R_SMALL_XHT
-, R_EDGE_CHAR
-, R_1IL_CONFLICT
-, R_POSTNN_1IL
-, R_REJ_CBLOB
-, R_MM_REJECT
-, R_BAD_REPETITION
-, R_POOR_MATCH
-, R_NOT_TESS_ACCEPTED
-, R_CONTAINS_BLANKS
-, R_BAD_PERMUTER
-, R_HYPHEN
-, R_DUBIOUS
-, R_NO_ALPHANUMS
-, R_MOSTLY_REJ
-, R_XHT_FIXUP
-, R_BAD_QUALITY
-, R_DOC_REJ
-, R_BLOCK_REJ
-, R_ROW_REJ
-, R_UNLV_REJ
-, R_NN_ACCEPT
-, R_HYPHEN_ACCEPT
-, R_MM_ACCEPT
-, R_QUALITY_ACCEPT
-, R_MINIMAL_REJ_ACCEPT
- const double tesseract::TanhTable [] const double tesseract::LogisticTable []
-
-
-
◆
-
-
-
-
-
- #define MAP_ACCEPT '1'
-
-
-
-
-
-
◆
-
-
-
-
-
- #define MAP_REJECT_PERM '0'
-
-
-
-
-
-
◆
-
-
-
-
-
- #define MAP_REJECT_POTENTIAL '3'
-
-
-
-
-
-
◆
-
-
-
-
-
- #define MAP_REJECT_TEMP '2'
-
-
-
-
-
-
-
◆
-
-
-
-
-Enumerator tesseract: src/ccstruct/rejctmap.h Source File
+tesseract: src/lstm/functions.cpp Source File
-
Go to the documentation of this file.
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
46 #href_anchor"code" href="a00647.html">errcode.h"
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
93 #define MAP_REJECT_PERM '0'
-
94 #define MAP_REJECT_TEMP '2'
-
95 #define MAP_REJECT_POTENTIAL '3'
-
-
-
-
-
-
-
-
-
-
-
-
-
-
109 bool rej_before_nn_accept();
-
110 bool rej_between_nn_and_mm();
-
111 bool rej_between_mm_and_quality_accept();
-
112 bool rej_between_quality_and_minimal_rej_accept();
-
113 bool rej_before_mm_accept();
-
114 bool rej_before_quality_accept();
-
-
-
-
-
-
-
121 flags1 = source.flags1;
-
122 flags2 = source.flags2;
-
-
-
-
126 const REJ & source) {
-
127 flags1 = source.flags1;
-
128 flags2 = source.flags2;
-
-
-
-
-
-
134 return flags1.
bit (rej_flag);
-
-
136 return flags2.
bit (rej_flag - 16);
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
/usr/share/doc/packages/tesseract-ocr-devel/html/a00155.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00155.html 2022-03-29 06:13:01.357591771 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00155.html 2022-03-29 06:13:01.353591756 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/ratngs.h File Reference
+
tesseract: src/lstm/lstmrecognizer.h File Reference
@@ -85,365 +85,45 @@
-
+
Enumerations
+
-
#include <cassert>
-
#include <cfloat>
-
#include "clst.h "
-
#href_anchor"el" href="a00665_source.html">elst.h"
-
#include "fontinfo.h "
-
#href_anchor"el" href="a00758_source.html">genericvector.h"
-
#include "matrix.h "
-
#href_anchor"el" href="a00617_source.html">unichar.h"
-
#include "unicharset.h "
-
#href_anchor"el" href="a00095_source.html">werd.h"
+
#include "ccutil.h "
+
#href_anchor"el" href="a00512_source.html">helpers.h"
+
#include "imagedata.h "
+
#href_anchor"el" href="a00725_source.html">matrix.h"
+
#include "network.h "
+
#href_anchor"el" href="a00209_source.html">networkscratch.h"
+
#include "params.h "
+
#href_anchor"el" href="a00074_source.html">recodebeam.h"
+
#include "series.h "
+
#href_anchor"el" href="a00395_source.html">strngs.h"
+
#include "unicharcompress.h "
Go to the source code of this file.
class &#href_anchor"memItemRight" valign="bottom">BLOB_CHOICE &#href_anchor"memitem:"> class WERD_CHOICE class &#href_anchor"memItemRight" valign="bottom">tesseract::LSTMRecognizer &#href_anchor"memberdecls">
namespace tesseract &#href_anchor"memberdecls">
- using BLOB_CHOICE_LIST_VECTOR = GenericVector < BLOB_CHOICE_LIST * >&#href_anchor"memberdecls">
enum BlobChoiceClassifier { BCC_STATIC_CLASSIFIER
-, BCC_ADAPTED_CLASSIFIER
-, BCC_SPECKLE_CLASSIFIER
-, BCC_AMBIG
-, BCC_FAKE
-enum tesseract::TrainingFlags { tesseract::TF_INT_MODE = 1
+, tesseract::TF_COMPRESS_UNICHARSET = 64
} enum PermuterType { NO_PERM
-, PUNC_PERM
-, TOP_CHOICE_PERM
-, LOWER_CASE_PERM
-, UPPER_CASE_PERM
-, NGRAM_PERM
-, NUMBER_PERM
-, USER_PATTERN_PERM
-, SYSTEM_DAWG_PERM
-, DOC_DAWG_PERM
-, USER_DAWG_PERM
-, FREQ_DAWG_PERM
-, COMPOUND_PERM
-, NUM_PERMUTER_TYPES
- enum tesseract::ScriptPos { tesseract::SP_NORMAL
-, tesseract::SP_SUBSCRIPT
-, tesseract::SP_SUPERSCRIPT
-, tesseract::SP_DROPCAP
- }
-BLOB_CHOICE *&#href_anchor"memItemRight" valign="bottom">FindMatchingChoice (UNICHAR_ID char_id, BLOB_CHOICE_LIST *bc_list) const char * tesseract::ScriptPosToString (enum ScriptPos script_pos) bool EqualIgnoringCaseAndTerminalPunct (const WERD_CHOICE &word1, const WERD_CHOICE &word2)&#href_anchor"memitem:a2de5b7d5a020bb9823091dbece4ea36e"> void print_ratings_list (const char *msg, BLOB_CHOICE_LIST *ratings, const UNICHARSET ¤t_unicharset)&#href_anchor"groupheader">Typedef Documentation
-◆ ◆
-
-
-Enumerator
-
-
Definition at line 43 of file ratngs.h .
-
43 {
-
-
-
-
-
-
49 };
-
-
-
-
-
-
-
-
◆
-
-
-Enumerator
-
-
Definition at line 232 of file ratngs.h .
-
232 {
-
-
-
-
-
/usr/share/doc/packages/tesseract-ocr-devel/html/a00155.js differs (ASCII text)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00155.js 2020-03-26 00:00:00.000000000 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00155.js 2020-03-26 00:00:00.000000000 +0000
@@ -1,39 +1,8 @@
var a00155 =
[
- [ "BLOB_CHOICE", "a02865.html", "a02865" ],
- [ "WERD_CHOICE", "a02869.html", "a02869" ],
- [ "BLOB_CHOICE_LIST_VECTOR", "a00155.html#aadc61f9ab58beb9c7171eed237f0f61d", null ],
- [ "BlobChoiceClassifier", "a00155.html#ac7c62d4383ee54bc9c16ef51a0ec3207", [
- [ "BCC_STATIC_CLASSIFIER", "a00155.html#ac7c62d4383ee54bc9c16ef51a0ec3207af99f6ba7d0068fe55ec1d462144c2417", null ],
- [ "BCC_ADAPTED_CLASSIFIER", "a00155.html#ac7c62d4383ee54bc9c16ef51a0ec3207a08697f2ff3b04a4ce9f4b869660e9f33", null ],
- [ "BCC_SPECKLE_CLASSIFIER", "a00155.html#ac7c62d4383ee54bc9c16ef51a0ec3207a807c445ba9b972d41a358a0a695076a1", null ],
- [ "BCC_AMBIG", "a00155.html#ac7c62d4383ee54bc9c16ef51a0ec3207a91ff206d2a86f2a3e9682ae4ba5bdd21", null ],
- [ "BCC_FAKE", "a00155.html#ac7c62d4383ee54bc9c16ef51a0ec3207a5d1ababd7ec62c799477a1ff0c33723f", null ]
- ] ],
- [ "PermuterType", "a00155.html#a18e2c75cefe9e5b78e8ce41aa5fa25bc", [
- [ "NO_PERM", "a00155.html#a18e2c75cefe9e5b78e8ce41aa5fa25bcafbce35860451a2bdc0a266cd83a4397f", null ],
- [ "PUNC_PERM", "a00155.html#a18e2c75cefe9e5b78e8ce41aa5fa25bcaa416e1922ec40e584a929afc70a90fe6", null ],
- [ "TOP_CHOICE_PERM", "a00155.html#a18e2c75cefe9e5b78e8ce41aa5fa25bca12d29cc139fe00da7420978c101d2889", null ],
- [ "LOWER_CASE_PERM", "a00155.html#a18e2c75cefe9e5b78e8ce41aa5fa25bcab997c3c17318b4549ee9255422c25f16", null ],
- [ "UPPER_CASE_PERM", "a00155.html#a18e2c75cefe9e5b78e8ce41aa5fa25bcad09d7ce85c0f3d35179e887e54f15b62", null ],
- [ "NGRAM_PERM", "a00155.html#a18e2c75cefe9e5b78e8ce41aa5fa25bca75b442d44a0439fa11444512c47576e4", null ],
- [ "NUMBER_PERM", "a00155.html#a18e2c75cefe9e5b78e8ce41aa5fa25bca809574e2067a5304fbd2279d869e9b24", null ],
- [ "USER_PATTERN_PERM", "a00155.html#a18e2c75cefe9e5b78e8ce41aa5fa25bca4004377f07488741fde22cbc8d2ae729", null ],
- [ "SYSTEM_DAWG_PERM", "a00155.html#a18e2c75cefe9e5b78e8ce41aa5fa25bca616bdbcaed7dd414823b303af5c8b1f1", null ],
- [ "DOC_DAWG_PERM", "a00155.html#a18e2c75cefe9e5b78e8ce41aa5fa25bca40f46ca413dfd53f7d7938c9fc305d04", null ],
- [ "USER_DAWG_PERM", "a00155.html#a18e2c75cefe9e5b78e8ce41aa5fa25bca5e293e54fa687876197992131e779d54", null ],
- [ "FREQ_DAWG_PERM", "a00155.html#a18e2c75cefe9e5b78e8ce41aa5fa25bca4fd9d3b025ecc5ea8f2248fea29543e4", null ],
- [ "COMPOUND_PERM", "a00155.html#a18e2c75cefe9e5b78e8ce41aa5fa25bcaea73d015098d9432d1be3c569db95bb5", null ],
- [ "NUM_PERMUTER_TYPES", "a00155.html#a18e2c75cefe9e5b78e8ce41aa5fa25bcae57ff54d1c91de8a7f32b8a5261e4d75", null ]
- ] ],
- [ "ScriptPos", "a00155.html#a496c25ad3b69eb36c48ae9cbf15f2fe2", [
- [ "SP_NORMAL", "a00155.html#a496c25ad3b69eb36c48ae9cbf15f2fe2a4aa5a5c8505a529a803f4a837d54c526", null ],
- [ "SP_SUBSCRIPT", "a00155.html#a496c25ad3b69eb36c48ae9cbf15f2fe2a0018c0645e510228c6701ff42d36bd01", null ],
- [ "SP_SUPERSCRIPT", "a00155.html#a496c25ad3b69eb36c48ae9cbf15f2fe2a6160fc6d2a9bba3b56e3f673f84b24b8", null ],
- [ "SP_DROPCAP", "a00155.html#a496c25ad3b69eb36c48ae9cbf15f2fe2a39a9b42f2bd03c3cb815dcf93b3d00d4", null ]
- ] ],
- [ "EqualIgnoringCaseAndTerminalPunct", "a00155.html#a98992670ff5a4da6695f318643d48ca4", null ],
- [ "FindMatchingChoice", "a00155.html#ae3be02b870f027a802148fc638810130", null ],
- [ "print_ratings_list", "a00155.html#a2de5b7d5a020bb9823091dbece4ea36e", null ],
- [ "ScriptPosToString", "a00155.html#a9a75fa73a4d69a48bdf5918c1949c976", null ]
+ [ "tesseract::LSTMRecognizer", "a03693.html", "a03693" ],
+ [ "TrainingFlags", "a00155.html#a0294eb0630c5f8dea13b0d8a58cc26e2", [
+ [ "TF_INT_MODE", "a00155.html#a0294eb0630c5f8dea13b0d8a58cc26e2a85f99732695115fa647f091760699816", null ],
+ [ "TF_COMPRESS_UNICHARSET", "a00155.html#a0294eb0630c5f8dea13b0d8a58cc26e2af91ba76cce718e575d71e5921515a6d3", null ]
+ ] ]
];
\ No newline at end of file
/usr/share/doc/packages/tesseract-ocr-devel/html/a00155_source.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00155_source.html 2022-03-29 06:13:01.413591985 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00155_source.html 2022-03-29 06:13:01.409591971 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/ratngs.h Source File
+
tesseract: src/lstm/lstmrecognizer.h Source File
@@ -82,805 +82,430 @@
-
Go to the documentation of this file.
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
20 #href_anchor"line"> 21 #define RATNGS_H
-
-
-
-
-
-
27 #href_anchor"code" href="a00665.html">elst.h"
-
28 #ifndef DISABLED_LEGACY_ENGINE
-
-
30 #href_anchor"comment">// ndef DISABLED_LEGACY_ENGINE
-
-
32 #href_anchor"code" href="a00098.html">matrix.h"
-
-
34 #href_anchor"code" href="a00737.html">unicharset.h"
-
-
-
-
-
+
Go to the documentation of this file.
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
19 #href_anchor"line"> 20 #define TESSERACT_LSTM_LSTMRECOGNIZER_H_
+
+
+
23 #href_anchor"code" href="a00512.html">helpers.h"
+
+
25 #href_anchor"code" href="a00725.html">matrix.h"
+
+
27 #href_anchor"code" href="a00209.html">networkscratch.h"
+
+
29 #href_anchor"code" href="a00074.html">recodebeam.h"
+
+
31 #href_anchor"code" href="a00395.html">strngs.h"
+
+
+
+
+
+
+
+
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
92 #ifndef DISABLED_LEGACY_ENGINE
-
-
-
-
-
-
98 int score1 = 0, score2 = 0;
-
-
-
101 for (
int f = 0; f < fonts_.
size (); ++f) {
-
102 if (fonts_[f].score > score1) {
-
-
104 fontinfo_id2_ = fontinfo_id_;
-
105 score1 = fonts_[f].score;
-
106 fontinfo_id_ = fonts_[f].fontinfo_id;
-
107 }
else if (fonts_[f].score > score2) {
-
108 score2 = fonts_[f].score;
-
109 fontinfo_id2_ = fonts_[f].fontinfo_id;
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
142 unichar_id_ = newunichar_id;
-
-
-
/usr/share/doc/packages/tesseract-ocr-devel/html/a00158.html differs (HTML document, ASCII text, with very long lines)
--- old//usr/share/doc/packages/tesseract-ocr-devel/html/a00158.html 2022-03-29 06:13:01.449592123 +0000
+++ new//usr/share/doc/packages/tesseract-ocr-devel/html/a00158.html 2022-03-29 06:13:01.437592077 +0000
@@ -5,7 +5,7 @@
-
tesseract: src/ccstruct/points.h File Reference
+
tesseract: src/lstm/reconfig.cpp File Reference
@@ -83,1126 +83,20 @@
+
-
#include <cmath>
-
#include <cstdio>
-
#include "elst.h "
-
#href_anchor"el" href="a00647_source.html">errcode.h"
-
#include "platform.h "
+
Go to the source code of this file.
-class &#href_anchor"memItemRight" valign="bottom">ICOORD &#href_anchor"mdescRight">integer coordinate More... class ICOORDELT &#href_anchor"memitem:"> class FCOORD &#href_anchor"memberdecls">
- ICOORD &#href_anchor"memItemRight" valign="bottom">operator! (const ICOORD &src)&#href_anchor"memitem:ad7ef7709c4047ce463a3ff3d61c0d965"> ICOORD &#href_anchor"memItemRight" valign="bottom">operator- (const ICOORD &src)&#href_anchor"memitem:afb03632aa6602c5193bf4af7964e6751"> ICOORD &#href_anchor"memItemRight" valign="bottom">operator+ (const ICOORD &op1, const ICOORD &op2)&#href_anchor"memitem:a17a3f5461e788ebd6fd26b81acd5a695"> ICOORD &&#href_anchor"memItemRight" valign="bottom">operator+= (ICOORD &op1, const ICOORD &op2)&#href_anchor"memitem:ab9a15dd7b99152f9e14b6cd8a28a4eb3"> ICOORD &#href_anchor"memItemRight" valign="bottom">operator- (const ICOORD &op1, const ICOORD &op2)&#href_anchor"memitem:a4e10181a10e49fcaad68ac9395823204"> ICOORD &&#href_anchor"memItemRight" valign="bottom">operator-= (ICOORD &op1, const ICOORD &op2)&#href_anchor"memitem:aa72e22db3f437cbe5c71f9ffb98fe99d"> int32_t operator% (const ICOORD &op1, const ICOORD &op2)&#href_anchor"memitem:a1afde4ca4db12bd495a9b77d0d6359fd"> int32_t operator* (const ICOORD &op1, const ICOORD &op2)&#href_anchor"memitem:ad9f927820402f297b30cb707d29298fb"> ICOORD &#href_anchor"memItemRight" valign="bottom">operator* (const ICOORD &op1, int16_t scale)&#href_anchor"memitem:a3e82ada467eb92dd5c68818747ea88fd"> ICOORD &#href_anchor"memItemRight" valign="bottom">operator* (int16_t scale, const ICOORD &op1)&#href_anchor"memitem:a21dbbedd5602a5dd743b3a82905a7adb"> ICOORD &&#href_anchor"memItemRight" valign="bottom">operator*= (ICOORD &op1, int16_t scale)&#href_anchor"memitem:a9b2ec486dd840f200952dbf566c742c3"> ICOORD &#href_anchor"memItemRight" valign="bottom">operator/ (const ICOORD &op1, int16_t scale)&#href_anchor"memitem:a4eb0f1ffba3ebeb350dfa1c0f2b8739d"> ICOORD &&#href_anchor"memItemRight" valign="bottom">operator/= (ICOORD &op1, int16_t scale)&#href_anchor"memitem:aa7c6e3433850394bff1cde53f51b70f3"> FCOORD &#href_anchor"memItemRight" valign="bottom">operator! (const FCOORD &src)&#href_anchor"memitem:ab4f0ee803a74deb221ffdb4bb27df2bd"> FCOORD &#href_anchor"memItemRight" valign="bottom">operator- (const FCOORD &src)&#href_anchor"memitem:a3407de919f1bbcde13608526c3449fcf"> FCOORD &#href_anchor"memItemRight" valign="bottom">operator+ (const FCOORD &op1, const FCOORD &op2)&#href_anchor"memitem:afa5552e421a657c14fa2a3efe0f77e9a"> FCOORD &&#href_anchor"memItemRight" valign="bottom">operator+= (FCOORD &op1, const FCOORD &op2)&#href_anchor"memitem:a9dd91936917a1b070c509f14ea5c6f6c"> FCOORD &#href_anchor"memItemRight" valign="bottom">operator- (const FCOORD &op1, const FCOORD &op2)&#href_anchor"memitem:a439557246ca09d7f14f677540e520d20"> FCOORD &&#href_anchor"memItemRight" valign="bottom">operator-= (FCOORD &op1, const FCOORD &op2)&#href_anchor"memitem:a0030744614a08aa765730942724339d9"> float operator% (const FCOORD &op1, const FCOORD &op2)&#href_anchor"memitem:ad82594734c3f7b48fcf30bfa2453f3c4"> float operator* (const FCOORD &op1, const FCOORD &op2)&#href_anchor"memitem:ae49d94e2d4773774716c0769f4514923"> FCOORD &#href_anchor"memItemRight" valign="bottom">operator* (const FCOORD &op1, float scale)&#href_anchor"memitem:aa9f8d3fa3aa83fc58bad4b31d2f7c920"> FCOORD &#href_anchor"memItemRight" valign="bottom">operator* (float scale, const FCOORD &op1)&#href_anchor"memitem:a0c5816a89875def646fb8d8939ffd274"> FCOORD &&#href_anchor"memItemRight" valign="bottom">operator*= (FCOORD &op1, float scale)&#href_anchor"memitem:aab2a4133b74ee9382c7e1fbcaefdbb6d"> FCOORD &#href_anchor"memItemRight" valign="bottom">operator/ (const FCOORD &op1, float scale)&#href_anchor"memitem:aff47bb3c78239e862b427a01a06daee4"> FCOORD &&#href_anchor"memItemRight" valign="bottom">operator/= (FCOORD &op1, float scale)&#href_anchor"groupheader">Function Documentation
-◆ [1/2]
1.9.3