tesseract  4.0.0-1-g2a2b
associate.cpp
Go to the documentation of this file.
1 // File: associate.cpp
3 // Description: Functions for scoring segmentation paths according to
4 // their character widths, gap widths and seam cuts.
5 // Author: Daria Antonova
6 // Created: Mon Mar 8 11:26:43 PDT 2010
7 //
8 // (C) Copyright 2010, Google Inc.
9 // Licensed under the Apache License, Version 2.0 (the "License");
10 // you may not use this file except in compliance with the License.
11 // You may obtain a copy of the License at
12 // http://www.apache.org/licenses/LICENSE-2.0
13 // Unless required by applicable law or agreed to in writing, software
14 // distributed under the License is distributed on an "AS IS" BASIS,
15 // WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
16 // See the License for the specific language governing permissions and
17 // limitations under the License.
18 //
20 
21 
22 #include <cmath>
23 #include <cstdio>
24 
25 #include "associate.h"
26 #include "normalis.h"
27 #include "pageres.h"
28 
29 namespace tesseract {
30 
32 const float AssociateUtils::kMinGap = 0.03f;
33 
34 void AssociateUtils::ComputeStats(int col, int row,
35  const AssociateStats *parent_stats,
36  int parent_path_length,
37  bool fixed_pitch,
38  float max_char_wh_ratio,
39  WERD_RES *word_res,
40  bool debug,
41  AssociateStats *stats) {
42  stats->Clear();
43 
44  ASSERT_HOST(word_res != nullptr);
45  if (word_res->blob_widths.empty()) {
46  return;
47  }
48  if (debug) {
49  tprintf("AssociateUtils::ComputeStats() for col=%d, row=%d%s\n",
50  col, row, fixed_pitch ? " (fixed pitch)" : "");
51  }
52  float normalizing_height = kBlnXHeight;
53  ROW* blob_row = word_res->blob_row;
54  // TODO(rays/daria) Can unicharset.script_has_xheight be useful here?
55  if (fixed_pitch && blob_row != nullptr) {
56  // For fixed pitch language like CJK, we use the full text height
57  // as the normalizing factor so we are not dependent on xheight
58  // calculation.
59  if (blob_row->body_size() > 0.0f) {
60  normalizing_height = word_res->denorm.y_scale() * blob_row->body_size();
61  } else {
62  normalizing_height = word_res->denorm.y_scale() *
63  (blob_row->x_height() + blob_row->ascenders());
64  }
65  if (debug) {
66  tprintf("normalizing height = %g (scale %g xheight %g ascenders %g)\n",
67  normalizing_height, word_res->denorm.y_scale(),
68  blob_row->x_height(), blob_row->ascenders());
69  }
70  }
71  float wh_ratio = word_res->GetBlobsWidth(col, row) / normalizing_height;
72  if (wh_ratio > max_char_wh_ratio) stats->bad_shape = true;
73  // Compute the gap sum for this shape. If there are only negative or only
74  // positive gaps, record their sum in stats->gap_sum. However, if there is
75  // a mixture, record only the sum of the positive gaps.
76  // TODO(antonova): explain fragment.
77  int negative_gap_sum = 0;
78  for (int c = col; c < row; ++c) {
79  int gap = word_res->GetBlobsGap(c);
80  (gap > 0) ? stats->gap_sum += gap : negative_gap_sum += gap;
81  }
82  if (stats->gap_sum == 0) stats->gap_sum = negative_gap_sum;
83  if (debug) {
84  tprintf("wh_ratio=%g (max_char_wh_ratio=%g) gap_sum=%d %s\n",
85  wh_ratio, max_char_wh_ratio, stats->gap_sum,
86  stats->bad_shape ? "bad_shape" : "");
87  }
88  // Compute shape_cost (for fixed pitch mode).
89  if (fixed_pitch) {
90  bool end_row = (row == (word_res->ratings->dimension() - 1));
91 
92  // Ensure that the blob has gaps on the left and the right sides
93  // (except for beginning and ending punctuation) and that there is
94  // no cutting through ink at the blob boundaries.
95  if (col > 0) {
96  float left_gap = word_res->GetBlobsGap(col - 1) / normalizing_height;
97  SEAM *left_seam = word_res->seam_array[col - 1];
98  if ((!end_row && left_gap < kMinGap) || left_seam->priority() > 0.0f) {
99  stats->bad_shape = true;
100  }
101  if (debug) {
102  tprintf("left_gap %g, left_seam %g %s\n", left_gap,
103  left_seam->priority(), stats->bad_shape ? "bad_shape" : "");
104  }
105  }
106  float right_gap = 0.0f;
107  if (!end_row) {
108  right_gap = word_res->GetBlobsGap(row) / normalizing_height;
109  SEAM *right_seam = word_res->seam_array[row];
110  if (right_gap < kMinGap || right_seam->priority() > 0.0f) {
111  stats->bad_shape = true;
112  if (right_gap < kMinGap) stats->bad_fixed_pitch_right_gap = true;
113  }
114  if (debug) {
115  tprintf("right_gap %g right_seam %g %s\n", right_gap,
116  right_seam->priority(), stats->bad_shape ? "bad_shape" : "");
117  }
118  }
119 
120  // Impose additional segmentation penalties if blob widths or gaps
121  // distribution don't fit a fixed-pitch model.
122  // Since we only know the widths and gaps of the path explored so far,
123  // the means and variances are computed for the path so far (not
124  // considering characters to the right of the last character on the path).
125  stats->full_wh_ratio = wh_ratio + right_gap;
126  if (parent_stats != nullptr) {
127  stats->full_wh_ratio_total =
128  (parent_stats->full_wh_ratio_total + stats->full_wh_ratio);
129  float mean =
130  stats->full_wh_ratio_total / static_cast<float>(parent_path_length+1);
131  stats->full_wh_ratio_var =
132  parent_stats->full_wh_ratio_var + pow(mean-stats->full_wh_ratio, 2);
133  } else {
134  stats->full_wh_ratio_total = stats->full_wh_ratio;
135  }
136  if (debug) {
137  tprintf("full_wh_ratio %g full_wh_ratio_total %g full_wh_ratio_var %g\n",
138  stats->full_wh_ratio, stats->full_wh_ratio_total,
139  stats->full_wh_ratio_var);
140  }
141 
142  stats->shape_cost =
143  FixedPitchWidthCost(wh_ratio, right_gap, end_row, max_char_wh_ratio);
144 
145  // For some reason Tesseract prefers to treat the whole CJ words
146  // as one blob when the initial segmentation is particularly bad.
147  // This hack is to avoid favoring such states.
148  if (col == 0 && end_row && wh_ratio > max_char_wh_ratio) {
149  stats->shape_cost += 10;
150  }
151  stats->shape_cost += stats->full_wh_ratio_var;
152  if (debug) tprintf("shape_cost %g\n", stats->shape_cost);
153  }
154 }
155 
156 float AssociateUtils::FixedPitchWidthCost(float norm_width,
157  float right_gap,
158  bool end_pos,
159  float max_char_wh_ratio) {
160  float cost = 0.0f;
161  if (norm_width > max_char_wh_ratio) cost += norm_width;
162  if (norm_width > kMaxFixedPitchCharAspectRatio)
163  cost += norm_width * norm_width; // extra penalty for merging CJK chars
164  // Penalize skinny blobs, except for punctuation in the last position.
165  if (norm_width+right_gap < 0.5f && !end_pos) {
166  cost += 1.0f - (norm_width + right_gap);
167  }
168  return cost;
169 }
170 
171 } // namespace tesseract
static void ComputeStats(int col, int row, const AssociateStats *parent_stats, int parent_path_length, bool fixed_pitch, float max_char_wh_ratio, WERD_RES *word_res, bool debug, AssociateStats *stats)
Definition: associate.cpp:34
GenericVector< int > blob_widths
Definition: pageres.h:219
int GetBlobsGap(int blob_index)
Definition: pageres.cpp:746
Definition: seam.h:44
float y_scale() const
Definition: normalis.h:270
const int kBlnXHeight
Definition: normalis.h:24
float body_size() const
Definition: ocrrow.h:73
int GetBlobsWidth(int start_blob, int last_blob)
Definition: pageres.cpp:736
float ascenders() const
Definition: ocrrow.h:82
float priority() const
Definition: seam.h:65
static const float kMaxFixedPitchCharAspectRatio
Definition: associate.h:69
DENORM denorm
Definition: pageres.h:204
float x_height() const
Definition: ocrrow.h:64
int dimension() const
Definition: matrix.h:533
Definition: ocrrow.h:36
bool empty() const
Definition: genericvector.h:90
DLLSYM void tprintf(const char *format,...)
Definition: tprintf.cpp:37
GenericVector< SEAM * > seam_array
Definition: pageres.h:217
static float FixedPitchWidthCost(float norm_width, float right_gap, bool end_pos, float max_char_wh_ratio)
Definition: associate.cpp:156
MATRIX * ratings
Definition: pageres.h:231
ROW * blob_row
Definition: pageres.h:200
static const float kMinGap
Definition: associate.h:70
#define ASSERT_HOST(x)
Definition: errcode.h:84