Mercurial > hgrepos > Python2 > PyMuPDF
annotate mupdf-source/thirdparty/tesseract/src/wordrec/associate.cpp @ 21:2f43e400f144
Provide an "all" target to build both the sdist and the wheel
| author | Franz Glasner <fzglas.hg@dom66.de> |
|---|---|
| date | Fri, 19 Sep 2025 10:28:53 +0200 |
| parents | b50eed0cc0ef |
| children |
| rev | line source |
|---|---|
|
2
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1 /////////////////////////////////////////////////////////////////////// |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2 // File: associate.cpp |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
3 // Description: Functions for scoring segmentation paths according to |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
4 // their character widths, gap widths and seam cuts. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
5 // Author: Daria Antonova |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
6 // Created: Mon Mar 8 11:26:43 PDT 2010 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
7 // |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
8 // (C) Copyright 2010, Google Inc. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
9 // Licensed under the Apache License, Version 2.0 (the "License"); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
10 // you may not use this file except in compliance with the License. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
11 // You may obtain a copy of the License at |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
12 // http://www.apache.org/licenses/LICENSE-2.0 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
13 // Unless required by applicable law or agreed to in writing, software |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
14 // distributed under the License is distributed on an "AS IS" BASIS, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
15 // WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
16 // See the License for the specific language governing permissions and |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
17 // limitations under the License. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
18 // |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
19 /////////////////////////////////////////////////////////////////////// |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
20 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
21 #include <cmath> |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
22 #include <cstdio> |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
23 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
24 #include "associate.h" |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
25 #include "normalis.h" |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
26 #include "pageres.h" |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
27 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
28 namespace tesseract { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
29 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
30 const float AssociateUtils::kMaxFixedPitchCharAspectRatio = 2.0f; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
31 const float AssociateUtils::kMinGap = 0.03f; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
32 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
33 void AssociateUtils::ComputeStats(int col, int row, const AssociateStats *parent_stats, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
34 int parent_path_length, bool fixed_pitch, float max_char_wh_ratio, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
35 WERD_RES *word_res, bool debug, AssociateStats *stats) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
36 stats->Clear(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
37 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
38 ASSERT_HOST(word_res != nullptr); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
39 if (word_res->blob_widths.empty()) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
40 return; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
41 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
42 if (debug) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
43 tprintf("AssociateUtils::ComputeStats() for col=%d, row=%d%s\n", col, row, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
44 fixed_pitch ? " (fixed pitch)" : ""); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
45 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
46 float normalizing_height = kBlnXHeight; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
47 ROW *blob_row = word_res->blob_row; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
48 // TODO(rays/daria) Can unicharset.script_has_xheight be useful here? |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
49 if (fixed_pitch && blob_row != nullptr) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
50 // For fixed pitch language like CJK, we use the full text height |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
51 // as the normalizing factor so we are not dependent on xheight |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
52 // calculation. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
53 if (blob_row->body_size() > 0.0f) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
54 normalizing_height = word_res->denorm.y_scale() * blob_row->body_size(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
55 } else { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
56 normalizing_height = |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
57 word_res->denorm.y_scale() * (blob_row->x_height() + blob_row->ascenders()); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
58 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
59 if (debug) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
60 tprintf("normalizing height = %g (scale %g xheight %g ascenders %g)\n", normalizing_height, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
61 word_res->denorm.y_scale(), blob_row->x_height(), blob_row->ascenders()); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
62 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
63 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
64 float wh_ratio = word_res->GetBlobsWidth(col, row) / normalizing_height; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
65 if (wh_ratio > max_char_wh_ratio) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
66 stats->bad_shape = true; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
67 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
68 // Compute the gap sum for this shape. If there are only negative or only |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
69 // positive gaps, record their sum in stats->gap_sum. However, if there is |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
70 // a mixture, record only the sum of the positive gaps. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
71 // TODO(antonova): explain fragment. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
72 int negative_gap_sum = 0; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
73 for (int c = col; c < row; ++c) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
74 int gap = word_res->GetBlobsGap(c); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
75 (gap > 0) ? stats->gap_sum += gap : negative_gap_sum += gap; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
76 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
77 if (stats->gap_sum == 0) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
78 stats->gap_sum = negative_gap_sum; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
79 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
80 if (debug) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
81 tprintf("wh_ratio=%g (max_char_wh_ratio=%g) gap_sum=%d %s\n", wh_ratio, max_char_wh_ratio, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
82 stats->gap_sum, stats->bad_shape ? "bad_shape" : ""); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
83 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
84 // Compute shape_cost (for fixed pitch mode). |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
85 if (fixed_pitch) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
86 bool end_row = (row == (word_res->ratings->dimension() - 1)); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
87 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
88 // Ensure that the blob has gaps on the left and the right sides |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
89 // (except for beginning and ending punctuation) and that there is |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
90 // no cutting through ink at the blob boundaries. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
91 if (col > 0) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
92 float left_gap = word_res->GetBlobsGap(col - 1) / normalizing_height; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
93 SEAM *left_seam = word_res->seam_array[col - 1]; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
94 if ((!end_row && left_gap < kMinGap) || left_seam->priority() > 0.0f) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
95 stats->bad_shape = true; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
96 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
97 if (debug) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
98 tprintf("left_gap %g, left_seam %g %s\n", left_gap, left_seam->priority(), |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
99 stats->bad_shape ? "bad_shape" : ""); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
100 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
101 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
102 float right_gap = 0.0f; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
103 if (!end_row) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
104 right_gap = word_res->GetBlobsGap(row) / normalizing_height; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
105 SEAM *right_seam = word_res->seam_array[row]; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
106 if (right_gap < kMinGap || right_seam->priority() > 0.0f) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
107 stats->bad_shape = true; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
108 if (right_gap < kMinGap) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
109 stats->bad_fixed_pitch_right_gap = true; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
110 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
111 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
112 if (debug) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
113 tprintf("right_gap %g right_seam %g %s\n", right_gap, right_seam->priority(), |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
114 stats->bad_shape ? "bad_shape" : ""); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
115 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
116 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
117 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
118 // Impose additional segmentation penalties if blob widths or gaps |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
119 // distribution don't fit a fixed-pitch model. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
120 // Since we only know the widths and gaps of the path explored so far, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
121 // the means and variances are computed for the path so far (not |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
122 // considering characters to the right of the last character on the path). |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
123 stats->full_wh_ratio = wh_ratio + right_gap; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
124 if (parent_stats != nullptr) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
125 stats->full_wh_ratio_total = (parent_stats->full_wh_ratio_total + stats->full_wh_ratio); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
126 float mean = stats->full_wh_ratio_total / static_cast<float>(parent_path_length + 1); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
127 stats->full_wh_ratio_var = |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
128 parent_stats->full_wh_ratio_var + pow(mean - stats->full_wh_ratio, 2); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
129 } else { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
130 stats->full_wh_ratio_total = stats->full_wh_ratio; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
131 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
132 if (debug) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
133 tprintf("full_wh_ratio %g full_wh_ratio_total %g full_wh_ratio_var %g\n", |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
134 stats->full_wh_ratio, stats->full_wh_ratio_total, stats->full_wh_ratio_var); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
135 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
136 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
137 stats->shape_cost = FixedPitchWidthCost(wh_ratio, right_gap, end_row, max_char_wh_ratio); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
138 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
139 // For some reason Tesseract prefers to treat the whole CJ words |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
140 // as one blob when the initial segmentation is particularly bad. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
141 // This hack is to avoid favoring such states. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
142 if (col == 0 && end_row && wh_ratio > max_char_wh_ratio) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
143 stats->shape_cost += 10; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
144 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
145 stats->shape_cost += stats->full_wh_ratio_var; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
146 if (debug) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
147 tprintf("shape_cost %g\n", stats->shape_cost); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
148 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
149 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
150 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
151 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
152 float AssociateUtils::FixedPitchWidthCost(float norm_width, float right_gap, bool end_pos, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
153 float max_char_wh_ratio) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
154 float cost = 0.0f; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
155 if (norm_width > max_char_wh_ratio) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
156 cost += norm_width; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
157 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
158 if (norm_width > kMaxFixedPitchCharAspectRatio) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
159 cost += norm_width * norm_width; // extra penalty for merging CJK chars |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
160 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
161 // Penalize skinny blobs, except for punctuation in the last position. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
162 if (norm_width + right_gap < 0.5f && !end_pos) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
163 cost += 1.0f - (norm_width + right_gap); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
164 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
165 return cost; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
166 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
167 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
168 } // namespace tesseract |
