tesseract  3.05.00
 All Classes Namespaces Files Functions Variables Typedefs Enumerations Enumerator Friends Macros Groups Pages
makerow.cpp
Go to the documentation of this file.
1 /**********************************************************************
2  * File: makerow.cpp (Formerly makerows.c)
3  * Description: Code to arrange blobs into rows of text.
4  * Author: Ray Smith
5  * Created: Mon Sep 21 14:34:48 BST 1992
6  *
7  * (C) Copyright 1992, Hewlett-Packard Ltd.
8  ** Licensed under the Apache License, Version 2.0 (the "License");
9  ** you may not use this file except in compliance with the License.
10  ** You may obtain a copy of the License at
11  ** http://www.apache.org/licenses/LICENSE-2.0
12  ** Unless required by applicable law or agreed to in writing, software
13  ** distributed under the License is distributed on an "AS IS" BASIS,
14  ** WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
15  ** See the License for the specific language governing permissions and
16  ** limitations under the License.
17  *
18  **********************************************************************/
19 
20 #ifdef __UNIX__
21 #include <assert.h>
22 #endif
23 #include "stderr.h"
24 #include "blobbox.h"
25 #include "ccstruct.h"
26 #include "detlinefit.h"
27 #include "statistc.h"
28 #include "drawtord.h"
29 #include "blkocc.h"
30 #include "sortflts.h"
31 #include "oldbasel.h"
32 #include "textord.h"
33 #include "tordmain.h"
34 #include "underlin.h"
35 #include "makerow.h"
36 #include "tprintf.h"
37 #include "tovars.h"
38 
39 // Include automatically generated configuration file if running autoconf.
40 #ifdef HAVE_CONFIG_H
41 #include "config_auto.h"
42 #endif
43 
44 BOOL_VAR(textord_heavy_nr, FALSE, "Vigorously remove noise");
45 BOOL_VAR(textord_show_initial_rows, FALSE, "Display row accumulation");
46 BOOL_VAR(textord_show_parallel_rows, FALSE, "Display page correlated rows");
47 BOOL_VAR(textord_show_expanded_rows, FALSE, "Display rows after expanding");
48 BOOL_VAR(textord_show_final_rows, FALSE, "Display rows after final fitting");
49 BOOL_VAR(textord_show_final_blobs, FALSE, "Display blob bounds after pre-ass");
50 BOOL_VAR(textord_test_landscape, FALSE, "Tests refer to land/port");
51 BOOL_VAR(textord_parallel_baselines, TRUE, "Force parallel baselines");
52 BOOL_VAR(textord_straight_baselines, FALSE, "Force straight baselines");
53 BOOL_VAR(textord_old_baselines, TRUE, "Use old baseline algorithm");
54 BOOL_VAR(textord_old_xheight, FALSE, "Use old xheight algorithm");
55 BOOL_VAR(textord_fix_xheight_bug, TRUE, "Use spline baseline");
56 BOOL_VAR(textord_fix_makerow_bug, TRUE, "Prevent multiple baselines");
57 BOOL_VAR(textord_debug_xheights, FALSE, "Test xheight algorithms");
58 BOOL_VAR(textord_biased_skewcalc, TRUE, "Bias skew estimates with line length");
59 BOOL_VAR(textord_interpolating_skew, TRUE, "Interpolate across gaps");
60 INT_VAR(textord_skewsmooth_offset, 4, "For smooth factor");
61 INT_VAR(textord_skewsmooth_offset2, 1, "For smooth factor");
62 INT_VAR(textord_test_x, -MAX_INT32, "coord of test pt");
63 INT_VAR(textord_test_y, -MAX_INT32, "coord of test pt");
64 INT_VAR(textord_min_blobs_in_row, 4, "Min blobs before gradient counted");
65 INT_VAR(textord_spline_minblobs, 8, "Min blobs in each spline segment");
66 INT_VAR(textord_spline_medianwin, 6, "Size of window for spline segmentation");
67 INT_VAR(textord_max_blob_overlaps, 4,
68  "Max number of blobs a big blob can overlap");
69 INT_VAR(textord_min_xheight, 10, "Min credible pixel xheight");
70 double_VAR(textord_spline_shift_fraction, 0.02,
71  "Fraction of line spacing for quad");
72 double_VAR(textord_spline_outlier_fraction, 0.1,
73  "Fraction of line spacing for outlier");
74 double_VAR(textord_skew_ile, 0.5, "Ile of gradients for page skew");
75 double_VAR(textord_skew_lag, 0.02, "Lag for skew on row accumulation");
76 double_VAR(textord_linespace_iqrlimit, 0.2, "Max iqr/median for linespace");
77 double_VAR(textord_width_limit, 8, "Max width of blobs to make rows");
78 double_VAR(textord_chop_width, 1.5, "Max width before chopping");
79 double_VAR(textord_expansion_factor, 1.0,
80  "Factor to expand rows by in expand_rows");
81 double_VAR(textord_overlap_x, 0.375, "Fraction of linespace for good overlap");
82 double_VAR(textord_minxh, 0.25, "fraction of linesize for min xheight");
83 double_VAR(textord_min_linesize, 1.25, "* blob height for initial linesize");
84 double_VAR(textord_excess_blobsize, 1.3,
85  "New row made if blob makes row this big");
86 double_VAR(textord_occupancy_threshold, 0.4, "Fraction of neighbourhood");
87 double_VAR(textord_underline_width, 2.0, "Multiple of line_size for underline");
88 double_VAR(textord_min_blob_height_fraction, 0.75,
89  "Min blob height/top to include blob top into xheight stats");
90 double_VAR(textord_xheight_mode_fraction, 0.4,
91  "Min pile height to make xheight");
92 double_VAR(textord_ascheight_mode_fraction, 0.08,
93  "Min pile height to make ascheight");
94 double_VAR(textord_descheight_mode_fraction, 0.08,
95  "Min pile height to make descheight");
96 double_VAR(textord_ascx_ratio_min, 1.25, "Min cap/xheight");
97 double_VAR(textord_ascx_ratio_max, 1.8, "Max cap/xheight");
98 double_VAR(textord_descx_ratio_min, 0.25, "Min desc/xheight");
99 double_VAR(textord_descx_ratio_max, 0.6, "Max desc/xheight");
100 double_VAR(textord_xheight_error_margin, 0.1, "Accepted variation");
101 INT_VAR(textord_lms_line_trials, 12, "Number of linew fits to do");
102 BOOL_VAR(textord_new_initial_xheight, TRUE, "Use test xheight mechanism");
103 BOOL_VAR(textord_debug_blob, FALSE, "Print test blob information");
104 
105 #define MAX_HEIGHT_MODES 12
106 
107 const int kMinLeaderCount = 5;
108 
109 // Factored-out helper to build a single row from a list of blobs.
110 // Returns the mean blob size.
111 static float MakeRowFromBlobs(float line_size,
112  BLOBNBOX_IT* blob_it, TO_ROW_IT* row_it) {
113  blob_it->sort(blob_x_order);
114  blob_it->move_to_first();
115  TO_ROW* row = NULL;
116  float total_size = 0.0f;
117  int blob_count = 0;
118  // Add all the blobs to a single TO_ROW.
119  for (; !blob_it->empty(); blob_it->forward()) {
120  BLOBNBOX* blob = blob_it->extract();
121  int top = blob->bounding_box().top();
122  int bottom = blob->bounding_box().bottom();
123  if (row == NULL) {
124  row = new TO_ROW(blob, top, bottom, line_size);
125  row_it->add_before_then_move(row);
126  } else {
127  row->add_blob(blob, top, bottom, line_size);
128  }
129  total_size += top - bottom;
130  ++blob_count;
131  }
132  return blob_count > 0 ? total_size / blob_count : total_size;
133 }
134 
135 // Helper to make a row using the children of a single blob.
136 // Returns the mean size of the blobs created.
137 float MakeRowFromSubBlobs(TO_BLOCK* block, C_BLOB* blob, TO_ROW_IT* row_it) {
138  // The blobs made from the children will go in the small_blobs list.
139  BLOBNBOX_IT bb_it(&block->small_blobs);
140  C_OUTLINE_IT ol_it(blob->out_list());
141  // Get the children.
142  ol_it.set_to_list(ol_it.data()->child());
143  if (ol_it.empty())
144  return 0.0f;
145  for (ol_it.mark_cycle_pt(); !ol_it.cycled_list(); ol_it.forward()) {
146  // Deep copy the child outline and use that to make a blob.
147  C_BLOB* blob = new C_BLOB(C_OUTLINE::deep_copy(ol_it.data()));
148  // Correct direction as needed.
149  blob->CheckInverseFlagAndDirection();
150  BLOBNBOX* bbox = new BLOBNBOX(blob);
151  bb_it.add_after_then_move(bbox);
152  }
153  // Now we can make a row from the blobs.
154  return MakeRowFromBlobs(block->line_size, &bb_it, row_it);
155 }
156 
164 float make_single_row(ICOORD page_tr, bool allow_sub_blobs,
165  TO_BLOCK* block, TO_BLOCK_LIST* blocks) {
166  BLOBNBOX_IT blob_it = &block->blobs;
167  TO_ROW_IT row_it = block->get_rows();
168 
169  // Include all the small blobs and large blobs.
170  blob_it.add_list_after(&block->small_blobs);
171  blob_it.add_list_after(&block->noise_blobs);
172  blob_it.add_list_after(&block->large_blobs);
173  if (block->blobs.singleton() && allow_sub_blobs) {
174  blob_it.move_to_first();
175  float size = MakeRowFromSubBlobs(block, blob_it.data()->cblob(), &row_it);
176  if (size > block->line_size)
177  block->line_size = size;
178  } else if (block->blobs.empty()) {
179  // Make a fake blob.
180  C_BLOB* blob = C_BLOB::FakeBlob(block->block->bounding_box());
181  // The blobnbox owns the blob.
182  BLOBNBOX* bblob = new BLOBNBOX(blob);
183  blob_it.add_after_then_move(bblob);
184  }
185  MakeRowFromBlobs(block->line_size, &blob_it, &row_it);
186  // Fit an LMS line to the rows.
187  for (row_it.mark_cycle_pt(); !row_it.cycled_list(); row_it.forward())
188  fit_lms_line(row_it.data());
189  float gradient;
190  float fit_error;
191  // Compute the skew based on the fitted line.
192  compute_page_skew(blocks, gradient, fit_error);
193  return gradient;
194 }
195 
201 float make_rows(ICOORD page_tr, TO_BLOCK_LIST *port_blocks) {
202  float port_m; // global skew
203  float port_err; // global noise
204  TO_BLOCK_IT block_it; // iterator
205 
206  block_it.set_to_list(port_blocks);
207  for (block_it.mark_cycle_pt(); !block_it.cycled_list();
208  block_it.forward())
209  make_initial_textrows(page_tr, block_it.data(), FCOORD(1.0f, 0.0f),
210  !(BOOL8) textord_test_landscape);
211  // compute globally
212  compute_page_skew(port_blocks, port_m, port_err);
213  block_it.set_to_list(port_blocks);
214  for (block_it.mark_cycle_pt(); !block_it.cycled_list(); block_it.forward()) {
215  cleanup_rows_making(page_tr, block_it.data(), port_m, FCOORD(1.0f, 0.0f),
216  block_it.data()->block->bounding_box().left(),
217  !(BOOL8)textord_test_landscape);
218  }
219  return port_m; // global skew
220 }
221 
227 void make_initial_textrows( //find lines
228  ICOORD page_tr,
229  TO_BLOCK *block, //block to do
230  FCOORD rotation, //for drawing
231  BOOL8 testing_on //correct orientation
232  ) {
233  TO_ROW_IT row_it = block->get_rows ();
234 
235 #ifndef GRAPHICS_DISABLED
236  ScrollView::Color colour; //of row
237 
238  if (textord_show_initial_rows && testing_on) {
239  if (to_win == NULL)
240  create_to_win(page_tr);
241  }
242 #endif
243  //guess skew
244  assign_blobs_to_rows (block, NULL, 0, TRUE, TRUE, textord_show_initial_rows && testing_on);
245  row_it.move_to_first ();
246  for (row_it.mark_cycle_pt (); !row_it.cycled_list (); row_it.forward ())
247  fit_lms_line (row_it.data ());
248 #ifndef GRAPHICS_DISABLED
249  if (textord_show_initial_rows && testing_on) {
250  colour = ScrollView::RED;
251  for (row_it.mark_cycle_pt (); !row_it.cycled_list (); row_it.forward ()) {
252  plot_to_row (row_it.data (), colour, rotation);
253  colour = (ScrollView::Color) (colour + 1);
254  if (colour > ScrollView::MAGENTA)
255  colour = ScrollView::RED;
256  }
257  }
258 #endif
259 }
260 
261 
267 void fit_lms_line(TO_ROW *row) {
268  float m, c; // fitted line
269  tesseract::DetLineFit lms;
270  BLOBNBOX_IT blob_it = row->blob_list();
271 
272  for (blob_it.mark_cycle_pt(); !blob_it.cycled_list(); blob_it.forward()) {
273  const TBOX& box = blob_it.data()->bounding_box();
274  lms.Add(ICOORD((box.left() + box.right()) / 2, box.bottom()));
275  }
276  double error = lms.Fit(&m, &c);
277  row->set_line(m, c, error);
278 }
279 
280 
287 void compute_page_skew( //get average gradient
288  TO_BLOCK_LIST *blocks, //list of blocks
289  float &page_m, //average gradient
290  float &page_err //average error
291  ) {
292  inT32 row_count; //total rows
293  inT32 blob_count; //total_blobs
294  inT32 row_err; //integer error
295  float *gradients; //of rows
296  float *errors; //of rows
297  inT32 row_index; //of total
298  TO_ROW *row; //current row
299  TO_BLOCK_IT block_it = blocks; //iterator
300  TO_ROW_IT row_it;
301 
302  row_count = 0;
303  blob_count = 0;
304  for (block_it.mark_cycle_pt (); !block_it.cycled_list ();
305  block_it.forward ()) {
306  POLY_BLOCK* pb = block_it.data()->block->poly_block();
307  if (pb != NULL && !pb->IsText())
308  continue; // Pretend non-text blocks don't exist.
309  row_count += block_it.data ()->get_rows ()->length ();
310  //count up rows
311  row_it.set_to_list (block_it.data ()->get_rows ());
312  for (row_it.mark_cycle_pt (); !row_it.cycled_list (); row_it.forward ())
313  blob_count += row_it.data ()->blob_list ()->length ();
314  }
315  if (row_count == 0) {
316  page_m = 0.0f;
317  page_err = 0.0f;
318  return;
319  }
320  gradients = (float *) alloc_mem (blob_count * sizeof (float));
321  //get mem
322  errors = (float *) alloc_mem (blob_count * sizeof (float));
323  if (gradients == NULL || errors == NULL)
324  MEMORY_OUT.error ("compute_page_skew", ABORT, NULL);
325 
326  row_index = 0;
327  for (block_it.mark_cycle_pt (); !block_it.cycled_list ();
328  block_it.forward ()) {
329  POLY_BLOCK* pb = block_it.data()->block->poly_block();
330  if (pb != NULL && !pb->IsText())
331  continue; // Pretend non-text blocks don't exist.
332  row_it.set_to_list (block_it.data ()->get_rows ());
333  for (row_it.mark_cycle_pt (); !row_it.cycled_list (); row_it.forward ()) {
334  row = row_it.data ();
335  blob_count = row->blob_list ()->length ();
336  row_err = (inT32) ceil (row->line_error ());
337  if (row_err <= 0)
338  row_err = 1;
339  if (textord_biased_skewcalc) {
340  blob_count /= row_err;
341  for (blob_count /= row_err; blob_count > 0; blob_count--) {
342  gradients[row_index] = row->line_m ();
343  errors[row_index] = row->line_error ();
344  row_index++;
345  }
346  }
347  else if (blob_count >= textord_min_blobs_in_row) {
348  //get gradient
349  gradients[row_index] = row->line_m ();
350  errors[row_index] = row->line_error ();
351  row_index++;
352  }
353  }
354  }
355  if (row_index == 0) {
356  //desperate
357  for (block_it.mark_cycle_pt (); !block_it.cycled_list ();
358  block_it.forward ()) {
359  POLY_BLOCK* pb = block_it.data()->block->poly_block();
360  if (pb != NULL && !pb->IsText())
361  continue; // Pretend non-text blocks don't exist.
362  row_it.set_to_list (block_it.data ()->get_rows ());
363  for (row_it.mark_cycle_pt (); !row_it.cycled_list ();
364  row_it.forward ()) {
365  row = row_it.data ();
366  gradients[row_index] = row->line_m ();
367  errors[row_index] = row->line_error ();
368  row_index++;
369  }
370  }
371  }
372  row_count = row_index;
373  row_index = choose_nth_item ((inT32) (row_count * textord_skew_ile),
374  gradients, row_count);
375  page_m = gradients[row_index];
376  row_index = choose_nth_item ((inT32) (row_count * textord_skew_ile),
377  errors, row_count);
378  page_err = errors[row_index];
379  free_mem(gradients);
380  free_mem(errors);
381 }
382 
383 const double kNoiseSize = 0.5; // Fraction of xheight.
384 const int kMinSize = 8; // Min pixels to be xheight.
385 
390 static bool dot_of_i(BLOBNBOX* dot, BLOBNBOX* i, TO_ROW* row) {
391  const TBOX& ibox = i->bounding_box();
392  const TBOX& dotbox = dot->bounding_box();
393 
394  // Must overlap horizontally by enough and be high enough.
395  int overlap = MIN(dotbox.right(), ibox.right()) -
396  MAX(dotbox.left(), ibox.left());
397  if (ibox.height() <= 2 * dotbox.height() ||
398  (overlap * 2 < ibox.width() && overlap < dotbox.width()))
399  return false;
400 
401  // If the i is tall and thin then it is good.
402  if (ibox.height() > ibox.width() * 2)
403  return true; // The i or ! must be tall and thin.
404 
405  // It might still be tall and thin, but it might be joined to something.
406  // So search the outline for a piece of large height close to the edges
407  // of the dot.
408  const double kHeightFraction = 0.6;
409  double target_height = MIN(dotbox.bottom(), ibox.top());
410  target_height -= row->line_m()*dotbox.left() + row->line_c();
411  target_height *= kHeightFraction;
412  int left_min = dotbox.left() - dotbox.width();
413  int middle = (dotbox.left() + dotbox.right())/2;
414  int right_max = dotbox.right() + dotbox.width();
415  int left_miny = 0;
416  int left_maxy = 0;
417  int right_miny = 0;
418  int right_maxy = 0;
419  bool found_left = false;
420  bool found_right = false;
421  bool in_left = false;
422  bool in_right = false;
423  C_BLOB* blob = i->cblob();
424  C_OUTLINE_IT o_it = blob->out_list();
425  for (o_it.mark_cycle_pt(); !o_it.cycled_list(); o_it.forward()) {
426  C_OUTLINE* outline = o_it.data();
427  int length = outline->pathlength();
428  ICOORD pos = outline->start_pos();
429  for (int step = 0; step < length; pos += outline->step(step++)) {
430  int x = pos.x();
431  int y = pos.y();
432  if (x >= left_min && x < middle && !found_left) {
433  // We are in the left part so find min and max y.
434  if (in_left) {
435  if (y > left_maxy) left_maxy = y;
436  if (y < left_miny) left_miny = y;
437  } else {
438  left_maxy = left_miny = y;
439  in_left = true;
440  }
441  } else if (in_left) {
442  // We just left the left so look for size.
443  if (left_maxy - left_miny > target_height) {
444  if (found_right)
445  return true;
446  found_left = true;
447  }
448  in_left = false;
449  }
450  if (x <= right_max && x > middle && !found_right) {
451  // We are in the right part so find min and max y.
452  if (in_right) {
453  if (y > right_maxy) right_maxy = y;
454  if (y < right_miny) right_miny = y;
455  } else {
456  right_maxy = right_miny = y;
457  in_right = true;
458  }
459  } else if (in_right) {
460  // We just left the right so look for size.
461  if (right_maxy - right_miny > target_height) {
462  if (found_left)
463  return true;
464  found_right = true;
465  }
466  in_right = false;
467  }
468  }
469  }
470  return false;
471 }
472 
473 void vigorous_noise_removal(TO_BLOCK* block) {
474  TO_ROW_IT row_it = block->get_rows ();
475  for (row_it.mark_cycle_pt (); !row_it.cycled_list (); row_it.forward ()) {
476  TO_ROW* row = row_it.data();
477  BLOBNBOX_IT b_it = row->blob_list();
478  // Estimate the xheight on the row.
479  int max_height = 0;
480  for (b_it.mark_cycle_pt(); !b_it.cycled_list(); b_it.forward()) {
481  BLOBNBOX* blob = b_it.data();
482  if (blob->bounding_box().height() > max_height)
483  max_height = blob->bounding_box().height();
484  }
485  STATS hstats(0, max_height + 1);
486  for (b_it.mark_cycle_pt(); !b_it.cycled_list(); b_it.forward()) {
487  BLOBNBOX* blob = b_it.data();
488  int height = blob->bounding_box().height();
489  if (height >= kMinSize)
490  hstats.add(blob->bounding_box().height(), 1);
491  }
492  float xheight = hstats.median();
493  // Delete small objects.
494  BLOBNBOX* prev = NULL;
495  for (b_it.mark_cycle_pt(); !b_it.cycled_list(); b_it.forward()) {
496  BLOBNBOX* blob = b_it.data();
497  const TBOX& box = blob->bounding_box();
498  if (box.height() < kNoiseSize * xheight) {
499  // Small so delete unless it looks like an i dot.
500  if (prev != NULL) {
501  if (dot_of_i(blob, prev, row))
502  continue; // Looks OK.
503  }
504  if (!b_it.at_last()) {
505  BLOBNBOX* next = b_it.data_relative(1);
506  if (dot_of_i(blob, next, row))
507  continue; // Looks OK.
508  }
509  // It might be noise so get rid of it.
510  delete blob->cblob();
511  delete b_it.extract();
512  } else {
513  prev = blob;
514  }
515  }
516  }
517 }
518 
524 void cleanup_rows_making( //find lines
525  ICOORD page_tr, //top right
526  TO_BLOCK *block, //block to do
527  float gradient, //gradient to fit
528  FCOORD rotation, //for drawing
529  inT32 block_edge, //edge of block
530  BOOL8 testing_on //correct orientation
531  ) {
532  //iterators
533  BLOBNBOX_IT blob_it = &block->blobs;
534  TO_ROW_IT row_it = block->get_rows ();
535 
536 #ifndef GRAPHICS_DISABLED
537  if (textord_show_parallel_rows && testing_on) {
538  if (to_win == NULL)
539  create_to_win(page_tr);
540  }
541 #endif
542  //get row coords
543  fit_parallel_rows(block,
544  gradient,
545  rotation,
546  block_edge,
547  textord_show_parallel_rows &&testing_on);
548  delete_non_dropout_rows(block,
549  gradient,
550  rotation,
551  block_edge,
552  textord_show_parallel_rows &&testing_on);
553  expand_rows(page_tr, block, gradient, rotation, block_edge, testing_on);
554  blob_it.set_to_list (&block->blobs);
555  row_it.set_to_list (block->get_rows ());
556  for (row_it.mark_cycle_pt (); !row_it.cycled_list (); row_it.forward ())
557  blob_it.add_list_after (row_it.data ()->blob_list ());
558  //give blobs back
559  assign_blobs_to_rows (block, &gradient, 1, FALSE, FALSE, FALSE);
560  //now new rows must be genuine
561  blob_it.set_to_list (&block->blobs);
562  blob_it.add_list_after (&block->large_blobs);
563  assign_blobs_to_rows (block, &gradient, 2, TRUE, TRUE, FALSE);
564  //safe to use big ones now
565  blob_it.set_to_list (&block->blobs);
566  //throw all blobs in
567  blob_it.add_list_after (&block->noise_blobs);
568  blob_it.add_list_after (&block->small_blobs);
569  assign_blobs_to_rows (block, &gradient, 3, FALSE, FALSE, FALSE);
570 }
571 
577 void delete_non_dropout_rows( //find lines
578  TO_BLOCK *block, //block to do
579  float gradient, //global skew
580  FCOORD rotation, //deskew vector
581  inT32 block_edge, //left edge
582  BOOL8 testing_on //correct orientation
583  ) {
584  TBOX block_box; //deskewed block
585  inT32 *deltas; //change in occupation
586  inT32 *occupation; //of pixel coords
587  inT32 max_y; //in block
588  inT32 min_y;
589  inT32 line_index; //of scan line
590  inT32 line_count; //no of scan lines
591  inT32 distance; //to drop-out
592  inT32 xleft; //of block
593  inT32 ybottom; //of block
594  TO_ROW *row; //current row
595  TO_ROW_IT row_it = block->get_rows ();
596  BLOBNBOX_IT blob_it = &block->blobs;
597 
598  if (row_it.length () == 0)
599  return; //empty block
600  block_box = deskew_block_coords (block, gradient);
601  xleft = block->block->bounding_box ().left ();
602  ybottom = block->block->bounding_box ().bottom ();
603  min_y = block_box.bottom () - 1;
604  max_y = block_box.top () + 1;
605  for (row_it.mark_cycle_pt (); !row_it.cycled_list (); row_it.forward ()) {
606  line_index = (inT32) floor (row_it.data ()->intercept ());
607  if (line_index <= min_y)
608  min_y = line_index - 1;
609  if (line_index >= max_y)
610  max_y = line_index + 1;
611  }
612  line_count = max_y - min_y + 1;
613  if (line_count <= 0)
614  return; //empty block
615  deltas = (inT32 *) alloc_mem (line_count * sizeof (inT32));
616  occupation = (inT32 *) alloc_mem (line_count * sizeof (inT32));
617  if (deltas == NULL || occupation == NULL)
618  MEMORY_OUT.error ("compute_line_spacing", ABORT, NULL);
619 
620  compute_line_occupation(block, gradient, min_y, max_y, occupation, deltas);
621  compute_occupation_threshold ((inT32)
622  ceil (block->line_spacing *
623  (tesseract::CCStruct::kDescenderFraction +
624  tesseract::CCStruct::kAscenderFraction)),
625  (inT32) ceil (block->line_spacing *
626  (tesseract::CCStruct::kXHeightFraction +
627  tesseract::CCStruct::kAscenderFraction)),
628  max_y - min_y + 1, occupation, deltas);
629 #ifndef GRAPHICS_DISABLED
630  if (testing_on) {
631  draw_occupation(xleft, ybottom, min_y, max_y, occupation, deltas);
632  }
633 #endif
634  compute_dropout_distances(occupation, deltas, line_count);
635  for (row_it.mark_cycle_pt (); !row_it.cycled_list (); row_it.forward ()) {
636  row = row_it.data ();
637  line_index = (inT32) floor (row->intercept ());
638  distance = deltas[line_index - min_y];
639  if (find_best_dropout_row (row, distance, block->line_spacing / 2,
640  line_index, &row_it, testing_on)) {
641 #ifndef GRAPHICS_DISABLED
642  if (testing_on)
643  plot_parallel_row(row, gradient, block_edge,
644  ScrollView::WHITE, rotation);
645 #endif
646  blob_it.add_list_after (row_it.data ()->blob_list ());
647  delete row_it.extract (); //too far away
648  }
649  }
650  for (row_it.mark_cycle_pt (); !row_it.cycled_list (); row_it.forward ()) {
651  blob_it.add_list_after (row_it.data ()->blob_list ());
652  }
653 
654  free_mem(deltas);
655  free_mem(occupation);
656 }
657 
658 
665 BOOL8 find_best_dropout_row( //find neighbours
666  TO_ROW *row, //row to test
667  inT32 distance, //dropout dist
668  float dist_limit, //threshold distance
669  inT32 line_index, //index of row
670  TO_ROW_IT *row_it, //current position
671  BOOL8 testing_on //correct orientation
672  ) {
673  inT32 next_index; // of neighbouring row
674  inT32 row_offset; //from current row
675  inT32 abs_dist; //absolute distance
676  inT8 row_inc; //increment to row_index
677  TO_ROW *next_row; //nextious row
678 
679  if (testing_on)
680  tprintf ("Row at %g(%g), dropout dist=%d,",
681  row->intercept (), row->parallel_c (), distance);
682  if (distance < 0) {
683  row_inc = 1;
684  abs_dist = -distance;
685  }
686  else {
687  row_inc = -1;
688  abs_dist = distance;
689  }
690  if (abs_dist > dist_limit) {
691  if (testing_on) {
692  tprintf (" too far - deleting\n");
693  }
694  return TRUE;
695  }
696  if ((distance < 0 && !row_it->at_last ())
697  || (distance >= 0 && !row_it->at_first ())) {
698  row_offset = row_inc;
699  do {
700  next_row = row_it->data_relative (row_offset);
701  next_index = (inT32) floor (next_row->intercept ());
702  if ((distance < 0
703  && next_index < line_index
704  && next_index > line_index + distance + distance)
705  || (distance >= 0
706  && next_index > line_index
707  && next_index < line_index + distance + distance)) {
708  if (testing_on) {
709  tprintf (" nearer neighbour (%d) at %g\n",
710  line_index + distance - next_index,
711  next_row->intercept ());
712  }
713  return TRUE; //other is nearer
714  }
715  else if (next_index == line_index
716  || next_index == line_index + distance + distance) {
717  if (row->believability () <= next_row->believability ()) {
718  if (testing_on) {
719  tprintf (" equal but more believable at %g (%g/%g)\n",
720  next_row->intercept (),
721  row->believability (),
722  next_row->believability ());
723  }
724  return TRUE; //other is more believable
725  }
726  }
727  row_offset += row_inc;
728  }
729  while ((next_index == line_index
730  || next_index == line_index + distance + distance)
731  && row_offset < row_it->length ());
732  if (testing_on)
733  tprintf (" keeping\n");
734  }
735  return FALSE;
736 }
737 
738 
745 TBOX deskew_block_coords( //block box
746  TO_BLOCK *block, //block to do
747  float gradient //global skew
748  ) {
749  TBOX result; //block bounds
750  TBOX blob_box; //of block
751  FCOORD rotation; //deskew vector
752  float length; //of gradient vector
753  TO_ROW_IT row_it = block->get_rows ();
754  TO_ROW *row; //current row
755  BLOBNBOX *blob; //current blob
756  BLOBNBOX_IT blob_it; //iterator
757 
758  length = sqrt (gradient * gradient + 1);
759  rotation = FCOORD (1 / length, -gradient / length);
760  for (row_it.mark_cycle_pt (); !row_it.cycled_list (); row_it.forward ()) {
761  row = row_it.data ();
762  blob_it.set_to_list (row->blob_list ());
763  for (blob_it.mark_cycle_pt (); !blob_it.cycled_list ();
764  blob_it.forward ()) {
765  blob = blob_it.data ();
766  blob_box = blob->bounding_box ();
767  blob_box.rotate (rotation);//de-skew it
768  result += blob_box;
769  }
770  }
771  return result;
772 }
773 
774 
781 void compute_line_occupation( //project blobs
782  TO_BLOCK *block, //block to do
783  float gradient, //global skew
784  inT32 min_y, //min coord in block
785  inT32 max_y, //in block
786  inT32 *occupation, //output projection
787  inT32 *deltas //derivative
788  ) {
789  inT32 line_count; //maxy-miny+1
790  inT32 line_index; //of scan line
791  int index; //array index for daft compilers
792  float top, bottom; //coords of blob
793  inT32 width; //of blob
794  TO_ROW *row; //current row
795  TO_ROW_IT row_it = block->get_rows ();
796  BLOBNBOX *blob; //current blob
797  BLOBNBOX_IT blob_it; //iterator
798  float length; //of skew vector
799  TBOX blob_box; //bounding box
800  FCOORD rotation; //inverse of skew
801 
802  line_count = max_y - min_y + 1;
803  length = sqrt (gradient * gradient + 1);
804  rotation = FCOORD (1 / length, -gradient / length);
805  for (line_index = 0; line_index < line_count; line_index++)
806  deltas[line_index] = 0;
807  for (row_it.mark_cycle_pt (); !row_it.cycled_list (); row_it.forward ()) {
808  row = row_it.data ();
809  blob_it.set_to_list (row->blob_list ());
810  for (blob_it.mark_cycle_pt (); !blob_it.cycled_list ();
811  blob_it.forward ()) {
812  blob = blob_it.data ();
813  blob_box = blob->bounding_box ();
814  blob_box.rotate (rotation);//de-skew it
815  top = blob_box.top ();
816  bottom = blob_box.bottom ();
817  width =
818  (inT32) floor ((FLOAT32) (blob_box.right () - blob_box.left ()));
819  if ((inT32) floor (bottom) < min_y
820  || (inT32) floor (bottom) - min_y >= line_count)
821  fprintf (stderr,
822  "Bad y coord of bottom, " INT32FORMAT "(" INT32FORMAT ","
823  INT32FORMAT ")\n", (inT32) floor (bottom), min_y, max_y);
824  //count transitions
825  index = (inT32) floor (bottom) - min_y;
826  deltas[index] += width;
827  if ((inT32) floor (top) < min_y
828  || (inT32) floor (top) - min_y >= line_count)
829  fprintf (stderr,
830  "Bad y coord of top, " INT32FORMAT "(" INT32FORMAT ","
831  INT32FORMAT ")\n", (inT32) floor (top), min_y, max_y);
832  index = (inT32) floor (top) - min_y;
833  deltas[index] -= width;
834  }
835  }
836  occupation[0] = deltas[0];
837  for (line_index = 1; line_index < line_count; line_index++)
838  occupation[line_index] = occupation[line_index - 1] + deltas[line_index];
839 }
840 
841 
847 void compute_occupation_threshold( //project blobs
848  inT32 low_window, //below result point
849  inT32 high_window, //above result point
850  inT32 line_count, //array sizes
851  inT32 *occupation, //input projection
852  inT32 *thresholds //output thresholds
853  ) {
854  inT32 line_index; //of thresholds line
855  inT32 low_index; //in occupation
856  inT32 high_index; //in occupation
857  inT32 sum; //current average
858  inT32 divisor; //to get thresholds
859  inT32 min_index; //of min occ
860  inT32 min_occ; //min in locality
861  inT32 test_index; //for finding min
862 
863  divisor =
864  (inT32) ceil ((low_window + high_window) / textord_occupancy_threshold);
865  if (low_window + high_window < line_count) {
866  for (sum = 0, high_index = 0; high_index < low_window; high_index++)
867  sum += occupation[high_index];
868  for (low_index = 0; low_index < high_window; low_index++, high_index++)
869  sum += occupation[high_index];
870  min_occ = occupation[0];
871  min_index = 0;
872  for (test_index = 1; test_index < high_index; test_index++) {
873  if (occupation[test_index] <= min_occ) {
874  min_occ = occupation[test_index];
875  min_index = test_index; //find min in region
876  }
877  }
878  for (line_index = 0; line_index < low_window; line_index++)
879  thresholds[line_index] = (sum - min_occ) / divisor + min_occ;
880  //same out to end
881  for (low_index = 0; high_index < line_count; low_index++, high_index++) {
882  sum -= occupation[low_index];
883  sum += occupation[high_index];
884  if (occupation[high_index] <= min_occ) {
885  //find min in region
886  min_occ = occupation[high_index];
887  min_index = high_index;
888  }
889  //lost min from region
890  if (min_index <= low_index) {
891  min_occ = occupation[low_index + 1];
892  min_index = low_index + 1;
893  for (test_index = low_index + 2; test_index <= high_index;
894  test_index++) {
895  if (occupation[test_index] <= min_occ) {
896  min_occ = occupation[test_index];
897  //find min in region
898  min_index = test_index;
899  }
900  }
901  }
902  thresholds[line_index++] = (sum - min_occ) / divisor + min_occ;
903  }
904  }
905  else {
906  min_occ = occupation[0];
907  min_index = 0;
908  for (sum = 0, low_index = 0; low_index < line_count; low_index++) {
909  if (occupation[low_index] < min_occ) {
910  min_occ = occupation[low_index];
911  min_index = low_index;
912  }
913  sum += occupation[low_index];
914  }
915  line_index = 0;
916  }
917  for (; line_index < line_count; line_index++)
918  thresholds[line_index] = (sum - min_occ) / divisor + min_occ;
919  //same out to end
920 }
921 
922 
928 void compute_dropout_distances( //project blobs
929  inT32 *occupation, //input projection
930  inT32 *thresholds, //output thresholds
931  inT32 line_count //array sizes
932  ) {
933  inT32 line_index; //of thresholds line
934  inT32 distance; //from prev dropout
935  inT32 next_dist; //to next dropout
936  inT32 back_index; //for back filling
937  inT32 prev_threshold; //before overwrite
938 
939  distance = -line_count;
940  line_index = 0;
941  do {
942  do {
943  distance--;
944  prev_threshold = thresholds[line_index];
945  //distance from prev
946  thresholds[line_index] = distance;
947  line_index++;
948  }
949  while (line_index < line_count
950  && (occupation[line_index] < thresholds[line_index]
951  || occupation[line_index - 1] >= prev_threshold));
952  if (line_index < line_count) {
953  back_index = line_index - 1;
954  next_dist = 1;
955  while (next_dist < -distance && back_index >= 0) {
956  thresholds[back_index] = next_dist;
957  back_index--;
958  next_dist++;
959  distance++;
960  }
961  distance = 1;
962  }
963  }
964  while (line_index < line_count);
965 }
966 
967 
975 void expand_rows( //find lines
976  ICOORD page_tr, //top right
977  TO_BLOCK *block, //block to do
978  float gradient, //gradient to fit
979  FCOORD rotation, //for drawing
980  inT32 block_edge, //edge of block
981  BOOL8 testing_on //correct orientation
982  ) {
983  BOOL8 swallowed_row; //eaten a neighbour
984  float y_max, y_min; //new row limits
985  float y_bottom, y_top; //allowed limits
986  TO_ROW *test_row; //next row
987  TO_ROW *row; //current row
988  //iterators
989  BLOBNBOX_IT blob_it = &block->blobs;
990  TO_ROW_IT row_it = block->get_rows ();
991 
992 #ifndef GRAPHICS_DISABLED
993  if (textord_show_expanded_rows && testing_on) {
994  if (to_win == NULL)
995  create_to_win(page_tr);
996  }
997 #endif
998 
999  adjust_row_limits(block); //shift min,max.
1000  if (textord_new_initial_xheight) {
1001  if (block->get_rows ()->length () == 0)
1002  return;
1003  compute_row_stats(block, textord_show_expanded_rows &&testing_on);
1004  }
1005  assign_blobs_to_rows (block, &gradient, 4, TRUE, FALSE, FALSE);
1006  //get real membership
1007  if (block->get_rows ()->length () == 0)
1008  return;
1009  fit_parallel_rows(block,
1010  gradient,
1011  rotation,
1012  block_edge,
1013  textord_show_expanded_rows &&testing_on);
1014  if (!textord_new_initial_xheight)
1015  compute_row_stats(block, textord_show_expanded_rows &&testing_on);
1016  row_it.move_to_last ();
1017  do {
1018  row = row_it.data ();
1019  y_max = row->max_y (); //get current limits
1020  y_min = row->min_y ();
1021  y_bottom = row->intercept () - block->line_size * textord_expansion_factor *
1022  tesseract::CCStruct::kDescenderFraction;
1023  y_top = row->intercept () + block->line_size * textord_expansion_factor *
1024  (tesseract::CCStruct::kXHeightFraction +
1025  tesseract::CCStruct::kAscenderFraction);
1026  if (y_min > y_bottom) { //expansion allowed
1027  if (textord_show_expanded_rows && testing_on)
1028  tprintf("Expanding bottom of row at %f from %f to %f\n",
1029  row->intercept(), y_min, y_bottom);
1030  //expandable
1031  swallowed_row = TRUE;
1032  while (swallowed_row && !row_it.at_last ()) {
1033  swallowed_row = FALSE;
1034  //get next one
1035  test_row = row_it.data_relative (1);
1036  //overlaps space
1037  if (test_row->max_y () > y_bottom) {
1038  if (test_row->min_y () > y_bottom) {
1039  if (textord_show_expanded_rows && testing_on)
1040  tprintf("Eating row below at %f\n", test_row->intercept());
1041  row_it.forward ();
1042 #ifndef GRAPHICS_DISABLED
1043  if (textord_show_expanded_rows && testing_on)
1044  plot_parallel_row(test_row,
1045  gradient,
1046  block_edge,
1047  ScrollView::WHITE,
1048  rotation);
1049 #endif
1050  blob_it.set_to_list (row->blob_list ());
1051  blob_it.add_list_after (test_row->blob_list ());
1052  //swallow complete row
1053  delete row_it.extract ();
1054  row_it.backward ();
1055  swallowed_row = TRUE;
1056  }
1057  else if (test_row->max_y () < y_min) {
1058  //shorter limit
1059  y_bottom = test_row->max_y ();
1060  if (textord_show_expanded_rows && testing_on)
1061  tprintf("Truncating limit to %f due to touching row at %f\n",
1062  y_bottom, test_row->intercept());
1063  }
1064  else {
1065  y_bottom = y_min; //can't expand it
1066  if (textord_show_expanded_rows && testing_on)
1067  tprintf("Not expanding limit beyond %f due to touching row at %f\n",
1068  y_bottom, test_row->intercept());
1069  }
1070  }
1071  }
1072  y_min = y_bottom; //expand it
1073  }
1074  if (y_max < y_top) { //expansion allowed
1075  if (textord_show_expanded_rows && testing_on)
1076  tprintf("Expanding top of row at %f from %f to %f\n",
1077  row->intercept(), y_max, y_top);
1078  swallowed_row = TRUE;
1079  while (swallowed_row && !row_it.at_first ()) {
1080  swallowed_row = FALSE;
1081  //get one above
1082  test_row = row_it.data_relative (-1);
1083  if (test_row->min_y () < y_top) {
1084  if (test_row->max_y () < y_top) {
1085  if (textord_show_expanded_rows && testing_on)
1086  tprintf("Eating row above at %f\n", test_row->intercept());
1087  row_it.backward ();
1088  blob_it.set_to_list (row->blob_list ());
1089 #ifndef GRAPHICS_DISABLED
1090  if (textord_show_expanded_rows && testing_on)
1091  plot_parallel_row(test_row,
1092  gradient,
1093  block_edge,
1094  ScrollView::WHITE,
1095  rotation);
1096 #endif
1097  blob_it.add_list_after (test_row->blob_list ());
1098  //swallow complete row
1099  delete row_it.extract ();
1100  row_it.forward ();
1101  swallowed_row = TRUE;
1102  }
1103  else if (test_row->min_y () < y_max) {
1104  //shorter limit
1105  y_top = test_row->min_y ();
1106  if (textord_show_expanded_rows && testing_on)
1107  tprintf("Truncating limit to %f due to touching row at %f\n",
1108  y_top, test_row->intercept());
1109  }
1110  else {
1111  y_top = y_max; //can't expand it
1112  if (textord_show_expanded_rows && testing_on)
1113  tprintf("Not expanding limit beyond %f due to touching row at %f\n",
1114  y_top, test_row->intercept());
1115  }
1116  }
1117  }
1118  y_max = y_top;
1119  }
1120  //new limits
1121  row->set_limits (y_min, y_max);
1122  row_it.backward ();
1123  }
1124  while (!row_it.at_last ());
1125 }
1126 
1127 
1133 void adjust_row_limits( //tidy limits
1134  TO_BLOCK *block //block to do
1135  ) {
1136  TO_ROW *row; //current row
1137  float size; //size of row
1138  float ymax; //top of row
1139  float ymin; //bottom of row
1140  TO_ROW_IT row_it = block->get_rows ();
1141 
1142  if (textord_show_expanded_rows)
1143  tprintf("Adjusting row limits for block(%d,%d)\n",
1144  block->block->bounding_box().left(),
1145  block->block->bounding_box().top());
1146  for (row_it.mark_cycle_pt (); !row_it.cycled_list (); row_it.forward ()) {
1147  row = row_it.data ();
1148  size = row->max_y () - row->min_y ();
1149  if (textord_show_expanded_rows)
1150  tprintf("Row at %f has min %f, max %f, size %f\n",
1151  row->intercept(), row->min_y(), row->max_y(), size);
1152  size /= tesseract::CCStruct::kXHeightFraction +
1153  tesseract::CCStruct::kAscenderFraction +
1154  tesseract::CCStruct::kDescenderFraction;
1155  ymax = size * (tesseract::CCStruct::kXHeightFraction +
1156  tesseract::CCStruct::kAscenderFraction);
1157  ymin = -size * tesseract::CCStruct::kDescenderFraction;
1158  row->set_limits (row->intercept () + ymin, row->intercept () + ymax);
1159  row->merged = FALSE;
1160  }
1161 }
1162 
1163 
1169 void compute_row_stats( //find lines
1170  TO_BLOCK *block, //block to do
1171  BOOL8 testing_on //correct orientation
1172  ) {
1173  inT32 row_index; //of median
1174  TO_ROW *row; //current row
1175  TO_ROW *prev_row; //previous row
1176  float iqr; //inter quartile range
1177  TO_ROW_IT row_it = block->get_rows ();
1178  //number of rows
1179  inT16 rowcount = row_it.length ();
1180  TO_ROW **rows; //for choose nth
1181 
1182  rows = (TO_ROW **) alloc_mem (rowcount * sizeof (TO_ROW *));
1183  if (rows == NULL)
1184  MEMORY_OUT.error ("compute_row_stats", ABORT, NULL);
1185  rowcount = 0;
1186  prev_row = NULL;
1187  row_it.move_to_last (); //start at bottom
1188  do {
1189  row = row_it.data ();
1190  if (prev_row != NULL) {
1191  rows[rowcount++] = prev_row;
1192  prev_row->spacing = row->intercept () - prev_row->intercept ();
1193  if (testing_on)
1194  tprintf ("Row at %g yields spacing of %g\n",
1195  row->intercept (), prev_row->spacing);
1196  }
1197  prev_row = row;
1198  row_it.backward ();
1199  }
1200  while (!row_it.at_last ());
1201  block->key_row = prev_row;
1202  block->baseline_offset =
1203  fmod (prev_row->parallel_c (), block->line_spacing);
1204  if (testing_on)
1205  tprintf ("Blob based spacing=(%g,%g), offset=%g",
1206  block->line_size, block->line_spacing, block->baseline_offset);
1207  if (rowcount > 0) {
1208  row_index = choose_nth_item (rowcount * 3 / 4, rows, rowcount,
1209  sizeof (TO_ROW *), row_spacing_order);
1210  iqr = rows[row_index]->spacing;
1211  row_index = choose_nth_item (rowcount / 4, rows, rowcount,
1212  sizeof (TO_ROW *), row_spacing_order);
1213  iqr -= rows[row_index]->spacing;
1214  row_index = choose_nth_item (rowcount / 2, rows, rowcount,
1215  sizeof (TO_ROW *), row_spacing_order);
1216  block->key_row = rows[row_index];
1217  if (testing_on)
1218  tprintf (" row based=%g(%g)", rows[row_index]->spacing, iqr);
1219  if (rowcount > 2
1220  && iqr < rows[row_index]->spacing * textord_linespace_iqrlimit) {
1221  if (!textord_new_initial_xheight) {
1222  if (rows[row_index]->spacing < block->line_spacing
1223  && rows[row_index]->spacing > block->line_size)
1224  //within range
1225  block->line_size = rows[row_index]->spacing;
1226  //spacing=size
1227  else if (rows[row_index]->spacing > block->line_spacing)
1228  block->line_size = block->line_spacing;
1229  //too big so use max
1230  }
1231  else {
1232  if (rows[row_index]->spacing < block->line_spacing)
1233  block->line_size = rows[row_index]->spacing;
1234  else
1235  block->line_size = block->line_spacing;
1236  //too big so use max
1237  }
1238  if (block->line_size < textord_min_xheight)
1239  block->line_size = (float) textord_min_xheight;
1240  block->line_spacing = rows[row_index]->spacing;
1241  block->max_blob_size =
1242  block->line_spacing * textord_excess_blobsize;
1243  }
1244  block->baseline_offset = fmod (rows[row_index]->intercept (),
1245  block->line_spacing);
1246  }
1247  if (testing_on)
1248  tprintf ("\nEstimate line size=%g, spacing=%g, offset=%g\n",
1249  block->line_size, block->line_spacing, block->baseline_offset);
1250  free_mem(rows);
1251 }
1252 
1253 
1283 namespace tesseract {
1284 void Textord::compute_block_xheight(TO_BLOCK *block, float gradient) {
1285  TO_ROW *row; // current row
1286  float asc_frac_xheight = CCStruct::kAscenderFraction /
1287  CCStruct::kXHeightFraction;
1288  float desc_frac_xheight = CCStruct::kDescenderFraction /
1289  CCStruct::kXHeightFraction;
1290  inT32 min_height, max_height; // limits on xheight
1291  TO_ROW_IT row_it = block->get_rows();
1292  if (row_it.empty()) return; // no rows
1293 
1294  // Compute the best guess of xheight of each row individually.
1295  // Use xheight and ascrise values of the rows where ascenders were found.
1296  get_min_max_xheight(block->line_size, &min_height, &max_height);
1297  STATS row_asc_xheights(min_height, max_height + 1);
1298  STATS row_asc_ascrise(static_cast<int>(min_height * asc_frac_xheight),
1299  static_cast<int>(max_height * asc_frac_xheight) + 1);
1300  int min_desc_height = static_cast<int>(min_height * desc_frac_xheight);
1301  int max_desc_height = static_cast<int>(max_height * desc_frac_xheight);
1302  STATS row_asc_descdrop(min_desc_height, max_desc_height + 1);
1303  STATS row_desc_xheights(min_height, max_height + 1);
1304  STATS row_desc_descdrop(min_desc_height, max_desc_height + 1);
1305  STATS row_cap_xheights(min_height, max_height + 1);
1306  STATS row_cap_floating_xheights(min_height, max_height + 1);
1307  for (row_it.mark_cycle_pt(); !row_it.cycled_list(); row_it.forward()) {
1308  row = row_it.data();
1309  // Compute the xheight of this row if it has not been computed before.
1310  if (row->xheight <= 0.0) {
1311  compute_row_xheight(row, block->block->classify_rotation(),
1312  gradient, block->line_size);
1313  }
1314  ROW_CATEGORY row_category = get_row_category(row);
1315  if (row_category == ROW_ASCENDERS_FOUND) {
1316  row_asc_xheights.add(static_cast<inT32>(row->xheight),
1317  row->xheight_evidence);
1318  row_asc_ascrise.add(static_cast<inT32>(row->ascrise),
1319  row->xheight_evidence);
1320  row_asc_descdrop.add(static_cast<inT32>(-row->descdrop),
1321  row->xheight_evidence);
1322  } else if (row_category == ROW_DESCENDERS_FOUND) {
1323  row_desc_xheights.add(static_cast<inT32>(row->xheight),
1324  row->xheight_evidence);
1325  row_desc_descdrop.add(static_cast<inT32>(-row->descdrop),
1326  row->xheight_evidence);
1327  } else if (row_category == ROW_UNKNOWN) {
1328  fill_heights(row, gradient, min_height, max_height,
1329  &row_cap_xheights, &row_cap_floating_xheights);
1330  }
1331  }
1332 
1333  float xheight = 0.0;
1334  float ascrise = 0.0;
1335  float descdrop = 0.0;
1336  // Compute our best guess of xheight of this block.
1337  if (row_asc_xheights.get_total() > 0) {
1338  // Determine xheight from rows where ascenders were found.
1339  xheight = row_asc_xheights.median();
1340  ascrise = row_asc_ascrise.median();
1341  descdrop = -row_asc_descdrop.median();
1342  } else if (row_desc_xheights.get_total() > 0) {
1343  // Determine xheight from rows where descenders were found.
1344  xheight = row_desc_xheights.median();
1345  descdrop = -row_desc_descdrop.median();
1346  } else if (row_cap_xheights.get_total() > 0) {
1347  // All the rows in the block were (a/de)scenderless.
1348  // Try to search for two modes in row_cap_heights that could
1349  // be the xheight and the capheight (e.g. some of the rows
1350  // were lowercase, but did not have enough (a/de)scenders.
1351  // If such two modes can not be found, this block is most
1352  // likely all caps (or all small caps, in which case the code
1353  // still works as intended).
1354  compute_xheight_from_modes(&row_cap_xheights, &row_cap_floating_xheights,
1355  textord_single_height_mode &&
1356  block->block->classify_rotation().y() == 0.0,
1357  min_height, max_height, &(xheight), &(ascrise));
1358  if (ascrise == 0) { // assume only caps in the whole block
1359  xheight = row_cap_xheights.median() * CCStruct::kXHeightCapRatio;
1360  }
1361  } else { // default block sizes
1362  xheight = block->line_size * CCStruct::kXHeightFraction;
1363  }
1364  // Correct xheight, ascrise and descdrop if necessary.
1365  bool corrected_xheight = false;
1366  if (xheight < textord_min_xheight) {
1367  xheight = static_cast<float>(textord_min_xheight);
1368  corrected_xheight = true;
1369  }
1370  if (corrected_xheight || ascrise <= 0.0) {
1371  ascrise = xheight * asc_frac_xheight;
1372  }
1373  if (corrected_xheight || descdrop >= 0.0) {
1374  descdrop = -(xheight * desc_frac_xheight);
1375  }
1376  block->xheight = xheight;
1377 
1378  if (textord_debug_xheights) {
1379  tprintf("Block average xheight=%.4f, ascrise=%.4f, descdrop=%.4f\n",
1380  xheight, ascrise, descdrop);
1381  }
1382  // Correct xheight, ascrise, descdrop of rows based on block averages.
1383  for (row_it.mark_cycle_pt(); !row_it.cycled_list(); row_it.forward()) {
1384  correct_row_xheight(row_it.data(), xheight, ascrise, descdrop);
1385  }
1386 }
1387 
1396 void Textord::compute_row_xheight(TO_ROW *row, // row to do
1397  const FCOORD& rotation,
1398  float gradient, // global skew
1399  int block_line_size) {
1400  // Find blobs representing repeated characters in rows and mark them.
1401  // This information is used for computing row xheight and at a later
1402  // stage when words are formed by make_words.
1403  if (!row->rep_chars_marked()) {
1404  mark_repeated_chars(row);
1405  }
1406 
1407  int min_height, max_height;
1408  get_min_max_xheight(block_line_size, &min_height, &max_height);
1409  STATS heights(min_height, max_height + 1);
1410  STATS floating_heights(min_height, max_height + 1);
1411  fill_heights(row, gradient, min_height, max_height,
1412  &heights, &floating_heights);
1413  row->ascrise = 0.0f;
1414  row->xheight = 0.0f;
1415  row->xheight_evidence =
1416  compute_xheight_from_modes(&heights, &floating_heights,
1417  textord_single_height_mode &&
1418  rotation.y() == 0.0,
1419  min_height, max_height,
1420  &(row->xheight), &(row->ascrise));
1421  row->descdrop = 0.0f;
1422  if (row->xheight > 0.0) {
1423  row->descdrop = static_cast<float>(
1424  compute_row_descdrop(row, gradient, row->xheight_evidence, &heights));
1425  }
1426 }
1427 
1428 } // namespace tesseract.
1429 
1436 void fill_heights(TO_ROW *row, float gradient, int min_height,
1437  int max_height, STATS *heights, STATS *floating_heights) {
1438  float xcentre; // centre of blob
1439  float top; // top y coord of blob
1440  float height; // height of blob
1441  BLOBNBOX *blob; // current blob
1442  int repeated_set;
1443  BLOBNBOX_IT blob_it = row->blob_list();
1444  if (blob_it.empty()) return; // no blobs in this row
1445  bool has_rep_chars =
1446  row->rep_chars_marked() && row->num_repeated_sets() > 0;
1447  do {
1448  blob = blob_it.data();
1449  if (!blob->joined_to_prev()) {
1450  xcentre = (blob->bounding_box().left() +
1451  blob->bounding_box().right()) / 2.0f;
1452  top = blob->bounding_box().top();
1453  height = blob->bounding_box().height();
1454  if (textord_fix_xheight_bug)
1455  top -= row->baseline.y(xcentre);
1456  else
1457  top -= gradient * xcentre + row->parallel_c();
1458  if (top >= min_height && top <= max_height) {
1459  heights->add(static_cast<inT32>(floor(top + 0.5)), 1);
1460  if (height / top < textord_min_blob_height_fraction) {
1461  floating_heights->add(static_cast<inT32>(floor(top + 0.5)), 1);
1462  }
1463  }
1464  }
1465  // Skip repeated chars, since they are likely to skew the height stats.
1466  if (has_rep_chars && blob->repeated_set() != 0) {
1467  repeated_set = blob->repeated_set();
1468  blob_it.forward();
1469  while (!blob_it.at_first() &&
1470  blob_it.data()->repeated_set() == repeated_set) {
1471  blob_it.forward();
1472  if (textord_debug_xheights)
1473  tprintf("Skipping repeated char when computing xheight\n");
1474  }
1475  } else {
1476  blob_it.forward();
1477  }
1478  } while (!blob_it.at_first());
1479 }
1480 
1497 int compute_xheight_from_modes(
1498  STATS *heights, STATS *floating_heights, bool cap_only, int min_height,
1499  int max_height, float *xheight, float *ascrise) {
1500  int blob_index = heights->mode(); // find mode
1501  int blob_count = heights->pile_count(blob_index); // get count of mode
1502  if (textord_debug_xheights) {
1503  tprintf("min_height=%d, max_height=%d, mode=%d, count=%d, total=%d\n",
1504  min_height, max_height, blob_index, blob_count,
1505  heights->get_total());
1506  heights->print();
1507  floating_heights->print();
1508  }
1509  if (blob_count == 0) return 0;
1510  int modes[MAX_HEIGHT_MODES]; // biggest piles
1511  bool in_best_pile = FALSE;
1512  int prev_size = -MAX_INT32;
1513  int best_count = 0;
1514  int mode_count = compute_height_modes(heights, min_height, max_height,
1515  modes, MAX_HEIGHT_MODES);
1516  if (cap_only && mode_count > 1)
1517  mode_count = 1;
1518  int x;
1519  if (textord_debug_xheights) {
1520  tprintf("found %d modes: ", mode_count);
1521  for (x = 0; x < mode_count; x++) tprintf("%d ", modes[x]);
1522  tprintf("\n");
1523  }
1524 
1525  for (x = 0; x < mode_count - 1; x++) {
1526  if (modes[x] != prev_size + 1)
1527  in_best_pile = FALSE; // had empty height
1528  int modes_x_count = heights->pile_count(modes[x]) -
1529  floating_heights->pile_count(modes[x]);
1530  if ((modes_x_count >= blob_count * textord_xheight_mode_fraction) &&
1531  (in_best_pile || modes_x_count > best_count)) {
1532  for (int asc = x + 1; asc < mode_count; asc++) {
1533  float ratio =
1534  static_cast<float>(modes[asc]) / static_cast<float>(modes[x]);
1535  if (textord_ascx_ratio_min < ratio &&
1536  ratio < textord_ascx_ratio_max &&
1537  (heights->pile_count(modes[asc]) >=
1538  blob_count * textord_ascheight_mode_fraction)) {
1539  if (modes_x_count > best_count) {
1540  in_best_pile = true;
1541  best_count = modes_x_count;
1542  }
1543  if (textord_debug_xheights) {
1544  tprintf("X=%d, asc=%d, count=%d, ratio=%g\n",
1545  modes[x], modes[asc]-modes[x], modes_x_count, ratio);
1546  }
1547  prev_size = modes[x];
1548  *xheight = static_cast<float>(modes[x]);
1549  *ascrise = static_cast<float>(modes[asc] - modes[x]);
1550  }
1551  }
1552  }
1553  }
1554  if (*xheight == 0) { // single mode
1555  // Remove counts of the "floating" blobs (the one whose height is too
1556  // small in relation to it's top end of the bounding box) from heights
1557  // before computing the single-mode xheight.
1558  // Restore the counts in heights after the mode is found, since
1559  // floating blobs might be useful for determining potential ascenders
1560  // in compute_row_descdrop().
1561  if (floating_heights->get_total() > 0) {
1562  for (x = min_height; x < max_height; ++x) {
1563  heights->add(x, -(floating_heights->pile_count(x)));
1564  }
1565  blob_index = heights->mode(); // find the modified mode
1566  for (x = min_height; x < max_height; ++x) {
1567  heights->add(x, floating_heights->pile_count(x));
1568  }
1569  }
1570  *xheight = static_cast<float>(blob_index);
1571  *ascrise = 0.0f;
1572  best_count = heights->pile_count(blob_index);
1573  if (textord_debug_xheights)
1574  tprintf("Single mode xheight set to %g\n", *xheight);
1575  } else if (textord_debug_xheights) {
1576  tprintf("Multi-mode xheight set to %g, asc=%g\n", *xheight, *ascrise);
1577  }
1578  return best_count;
1579 }
1580 
1593 inT32 compute_row_descdrop(TO_ROW *row, float gradient,
1594  int xheight_blob_count, STATS *asc_heights) {
1595  // Count how many potential ascenders are in this row.
1596  int i_min = asc_heights->min_bucket();
1597  if ((i_min / row->xheight) < textord_ascx_ratio_min) {
1598  i_min = static_cast<int>(
1599  floor(row->xheight * textord_ascx_ratio_min + 0.5));
1600  }
1601  int i_max = asc_heights->max_bucket();
1602  if ((i_max / row->xheight) > textord_ascx_ratio_max) {
1603  i_max = static_cast<int>(floor(row->xheight * textord_ascx_ratio_max));
1604  }
1605  int num_potential_asc = 0;
1606  for (int i = i_min; i <= i_max; ++i) {
1607  num_potential_asc += asc_heights->pile_count(i);
1608  }
1609  inT32 min_height =
1610  static_cast<inT32>(floor(row->xheight * textord_descx_ratio_min + 0.5));
1611  inT32 max_height =
1612  static_cast<inT32>(floor(row->xheight * textord_descx_ratio_max));
1613  float xcentre; // centre of blob
1614  float height; // height of blob
1615  BLOBNBOX_IT blob_it = row->blob_list();
1616  BLOBNBOX *blob; // current blob
1617  STATS heights (min_height, max_height + 1);
1618  for (blob_it.mark_cycle_pt(); !blob_it.cycled_list(); blob_it.forward()) {
1619  blob = blob_it.data();
1620  if (!blob->joined_to_prev()) {
1621  xcentre = (blob->bounding_box().left() +
1622  blob->bounding_box().right()) / 2.0f;
1623  height = (gradient * xcentre + row->parallel_c() -
1624  blob->bounding_box().bottom());
1625  if (height >= min_height && height <= max_height)
1626  heights.add(static_cast<int>(floor(height + 0.5)), 1);
1627  }
1628  }
1629  int blob_index = heights.mode(); // find mode
1630  int blob_count = heights.pile_count(blob_index); // get count of mode
1631  float total_fraction =
1632  (textord_descheight_mode_fraction + textord_ascheight_mode_fraction);
1633  if (static_cast<float>(blob_count + num_potential_asc) <
1634  xheight_blob_count * total_fraction) {
1635  blob_count = 0;
1636  }
1637  int descdrop = blob_count > 0 ? -blob_index : 0;
1638  if (textord_debug_xheights) {
1639  tprintf("Descdrop: %d (potential ascenders %d, descenders %d)\n",
1640  descdrop, num_potential_asc, blob_count);
1641  heights.print();
1642  }
1643  return descdrop;
1644 }
1645 
1646 
1653 inT32 compute_height_modes(STATS *heights, // stats to search
1654  inT32 min_height, // bottom of range
1655  inT32 max_height, // top of range
1656  inT32 *modes, // output array
1657  inT32 maxmodes) { // size of modes
1658  inT32 pile_count; // no in source pile
1659  inT32 src_count; // no of source entries
1660  inT32 src_index; // current entry
1661  inT32 least_count; // height of smalllest
1662  inT32 least_index; // index of least
1663  inT32 dest_count; // index in modes
1664 
1665  src_count = max_height + 1 - min_height;
1666  dest_count = 0;
1667  least_count = MAX_INT32;
1668  least_index = -1;
1669  for (src_index = 0; src_index < src_count; src_index++) {
1670  pile_count = heights->pile_count(min_height + src_index);
1671  if (pile_count > 0) {
1672  if (dest_count < maxmodes) {
1673  if (pile_count < least_count) {
1674  // find smallest in array
1675  least_count = pile_count;
1676  least_index = dest_count;
1677  }
1678  modes[dest_count++] = min_height + src_index;
1679  } else if (pile_count >= least_count) {
1680  while (least_index < maxmodes - 1) {
1681  modes[least_index] = modes[least_index + 1];
1682  // shuffle up
1683  least_index++;
1684  }
1685  // new one on end
1686  modes[maxmodes - 1] = min_height + src_index;
1687  if (pile_count == least_count) {
1688  // new smallest
1689  least_index = maxmodes - 1;
1690  } else {
1691  least_count = heights->pile_count(modes[0]);
1692  least_index = 0;
1693  for (dest_count = 1; dest_count < maxmodes; dest_count++) {
1694  pile_count = heights->pile_count(modes[dest_count]);
1695  if (pile_count < least_count) {
1696  // find smallest
1697  least_count = pile_count;
1698  least_index = dest_count;
1699  }
1700  }
1701  }
1702  }
1703  }
1704  }
1705  return dest_count;
1706 }
1707 
1708 
1715 void correct_row_xheight(TO_ROW *row, float xheight,
1716  float ascrise, float descdrop) {
1717  ROW_CATEGORY row_category = get_row_category(row);
1718  if (textord_debug_xheights) {
1719  tprintf("correcting row xheight: row->xheight %.4f"
1720  ", row->acrise %.4f row->descdrop %.4f\n",
1721  row->xheight, row->ascrise, row->descdrop);
1722  }
1723  bool normal_xheight =
1724  within_error_margin(row->xheight, xheight, textord_xheight_error_margin);
1725  bool cap_xheight =
1726  within_error_margin(row->xheight, xheight + ascrise,
1727  textord_xheight_error_margin);
1728  // Use the average xheight/ascrise for the following cases:
1729  // -- the xheight of the row could not be determined at all
1730  // -- the row has descenders (e.g. "many groups", "ISBN 12345 p.3")
1731  // and its xheight is close to either cap height or average xheight
1732  // -- the row does not have ascenders or descenders, but its xheight
1733  // is close to the average block xheight (e.g. row with "www.mmm.com")
1734  if (row_category == ROW_ASCENDERS_FOUND) {
1735  if (row->descdrop >= 0.0) {
1736  row->descdrop = row->xheight * (descdrop / xheight);
1737  }
1738  } else if (row_category == ROW_INVALID ||
1739  (row_category == ROW_DESCENDERS_FOUND &&
1740  (normal_xheight || cap_xheight)) ||
1741  (row_category == ROW_UNKNOWN && normal_xheight)) {
1742  if (textord_debug_xheights) tprintf("using average xheight\n");
1743  row->xheight = xheight;
1744  row->ascrise = ascrise;
1745  row->descdrop = descdrop;
1746  } else if (row_category == ROW_DESCENDERS_FOUND) {
1747  // Assume this is a row with mostly lowercase letters and it's xheight
1748  // is computed correctly (unfortunately there is no way to distinguish
1749  // this from the case when descenders are found, but the most common
1750  // height is capheight).
1751  if (textord_debug_xheights) tprintf("lowercase, corrected ascrise\n");
1752  row->ascrise = row->xheight * (ascrise / xheight);
1753  } else if (row_category == ROW_UNKNOWN) {
1754  // Otherwise assume this row is an all-caps or small-caps row
1755  // and adjust xheight and ascrise of the row.
1756 
1757  row->all_caps = true;
1758  if (cap_xheight) { // regular all caps
1759  if (textord_debug_xheights) tprintf("all caps\n");
1760  row->xheight = xheight;
1761  row->ascrise = ascrise;
1762  row->descdrop = descdrop;
1763  } else { // small caps or caps with an odd xheight
1764  if (textord_debug_xheights) {
1765  if (row->xheight < xheight + ascrise && row->xheight > xheight) {
1766  tprintf("small caps\n");
1767  } else {
1768  tprintf("all caps with irregular xheight\n");
1769  }
1770  }
1771  row->ascrise = row->xheight * (ascrise / (xheight + ascrise));
1772  row->xheight -= row->ascrise;
1773  row->descdrop = row->xheight * (descdrop / xheight);
1774  }
1775  }
1776  if (textord_debug_xheights) {
1777  tprintf("corrected row->xheight = %.4f, row->acrise = %.4f, row->descdrop"
1778  " = %.4f\n", row->xheight, row->ascrise, row->descdrop);
1779  }
1780 }
1781 
1782 static int CountOverlaps(const TBOX& box, int min_height,
1783  BLOBNBOX_LIST* blobs) {
1784  int overlaps = 0;
1785  BLOBNBOX_IT blob_it(blobs);
1786  for (blob_it.mark_cycle_pt(); !blob_it.cycled_list(); blob_it.forward()) {
1787  BLOBNBOX* blob = blob_it.data();
1788  const TBOX &blob_box = blob->bounding_box();
1789  if (blob_box.height() >= min_height && box.major_overlap(blob_box)) {
1790  ++overlaps;
1791  }
1792  }
1793  return overlaps;
1794 }
1795 
1802 void separate_underlines(TO_BLOCK *block, // block to do
1803  float gradient, // skew angle
1804  FCOORD rotation, // inverse landscape
1805  BOOL8 testing_on) { // correct orientation
1806  BLOBNBOX *blob; // current blob
1807  C_BLOB *rotated_blob; // rotated blob
1808  TO_ROW *row; // current row
1809  float length; // of g_vec
1810  TBOX blob_box;
1811  FCOORD blob_rotation; // inverse of rotation
1812  FCOORD g_vec; // skew rotation
1813  BLOBNBOX_IT blob_it; // iterator
1814  // iterator
1815  BLOBNBOX_IT under_it = &block->underlines;
1816  BLOBNBOX_IT large_it = &block->large_blobs;
1817  TO_ROW_IT row_it = block->get_rows();
1818  int min_blob_height = static_cast<int>(textord_min_blob_height_fraction *
1819  block->line_size + 0.5);
1820 
1821  // length of vector
1822  length = sqrt(1 + gradient * gradient);
1823  g_vec = FCOORD(1 / length, -gradient / length);
1824  blob_rotation = FCOORD(rotation.x(), -rotation.y());
1825  blob_rotation.rotate(g_vec); // undoing everything
1826  for (row_it.mark_cycle_pt(); !row_it.cycled_list(); row_it.forward()) {
1827  row = row_it.data();
1828  // get blobs
1829  blob_it.set_to_list(row->blob_list());
1830  for (blob_it.mark_cycle_pt(); !blob_it.cycled_list();
1831  blob_it.forward()) {
1832  blob = blob_it.data();
1833  blob_box = blob->bounding_box();
1834  if (blob_box.width() > block->line_size * textord_underline_width) {
1835  ASSERT_HOST(blob->cblob() != NULL);
1836  rotated_blob = crotate_cblob (blob->cblob(),
1837  blob_rotation);
1838  if (test_underline(
1839  testing_on && textord_show_final_rows,
1840  rotated_blob, static_cast<inT16>(row->intercept()),
1841  static_cast<inT16>(
1842  block->line_size *
1843  (tesseract::CCStruct::kXHeightFraction +
1844  tesseract::CCStruct::kAscenderFraction / 2.0f)))) {
1845  under_it.add_after_then_move(blob_it.extract());
1846  if (testing_on && textord_show_final_rows) {
1847  tprintf("Underlined blob at:");
1848  rotated_blob->bounding_box().print();
1849  tprintf("Was:");
1850  blob_box.print();
1851  }
1852  } else if (CountOverlaps(blob->bounding_box(), min_blob_height,
1853  row->blob_list()) >
1854  textord_max_blob_overlaps) {
1855  large_it.add_after_then_move(blob_it.extract());
1856  if (testing_on && textord_show_final_rows) {
1857  tprintf("Large blob overlaps %d blobs at:",
1858  CountOverlaps(blob_box, min_blob_height,
1859  row->blob_list()));
1860  blob_box.print();
1861  }
1862  }
1863  delete rotated_blob;
1864  }
1865  }
1866  }
1867 }
1868 
1869 
1875 void pre_associate_blobs( //make rough chars
1876  ICOORD page_tr, //top right
1877  TO_BLOCK *block, //block to do
1878  FCOORD rotation, //inverse landscape
1879  BOOL8 testing_on //correct orientation
1880  ) {
1881 #ifndef GRAPHICS_DISABLED
1882  ScrollView::Color colour; //of boxes
1883 #endif
1884  BLOBNBOX *blob; //current blob
1885  BLOBNBOX *nextblob; //next in list
1886  TBOX blob_box;
1887  FCOORD blob_rotation; //inverse of rotation
1888  BLOBNBOX_IT blob_it; //iterator
1889  BLOBNBOX_IT start_it; //iterator
1890  TO_ROW_IT row_it = block->get_rows ();
1891 
1892 #ifndef GRAPHICS_DISABLED
1893  colour = ScrollView::RED;
1894 #endif
1895 
1896  blob_rotation = FCOORD (rotation.x (), -rotation.y ());
1897  for (row_it.mark_cycle_pt (); !row_it.cycled_list (); row_it.forward ()) {
1898  //get blobs
1899  blob_it.set_to_list (row_it.data ()->blob_list ());
1900  for (blob_it.mark_cycle_pt (); !blob_it.cycled_list ();
1901  blob_it.forward ()) {
1902  blob = blob_it.data ();
1903  blob_box = blob->bounding_box ();
1904  start_it = blob_it; //save start point
1905  // if (testing_on && textord_show_final_blobs)
1906  // {
1907  // tprintf("Blob at (%d,%d)->(%d,%d), addr=%x, count=%d\n",
1908  // blob_box.left(),blob_box.bottom(),
1909  // blob_box.right(),blob_box.top(),
1910  // (void*)blob,blob_it.length());
1911  // }
1912  bool overlap;
1913  do {
1914  overlap = false;
1915  if (!blob_it.at_last ()) {
1916  nextblob = blob_it.data_relative(1);
1917  overlap = blob_box.major_x_overlap(nextblob->bounding_box());
1918  if (overlap) {
1919  blob->merge(nextblob); // merge new blob
1920  blob_box = blob->bounding_box(); // get bigger box
1921  blob_it.forward();
1922  }
1923  }
1924  }
1925  while (overlap);
1926  blob->chop (&start_it, &blob_it,
1927  blob_rotation,
1928  block->line_size * tesseract::CCStruct::kXHeightFraction *
1929  textord_chop_width);
1930  //attempt chop
1931  }
1932 #ifndef GRAPHICS_DISABLED
1933  if (testing_on && textord_show_final_blobs) {
1934  if (to_win == NULL)
1935  create_to_win(page_tr);
1936  to_win->Pen(colour);
1937  for (blob_it.mark_cycle_pt (); !blob_it.cycled_list ();
1938  blob_it.forward ()) {
1939  blob = blob_it.data ();
1940  blob_box = blob->bounding_box ();
1941  blob_box.rotate (rotation);
1942  if (!blob->joined_to_prev ()) {
1943  to_win->Rectangle (blob_box.left (), blob_box.bottom (),
1944  blob_box.right (), blob_box.top ());
1945  }
1946  }
1947  colour = (ScrollView::Color) (colour + 1);
1948  if (colour > ScrollView::MAGENTA)
1949  colour = ScrollView::RED;
1950  }
1951 #endif
1952  }
1953 }
1954 
1955 
1961 void fit_parallel_rows( //find lines
1962  TO_BLOCK *block, //block to do
1963  float gradient, //gradient to fit
1964  FCOORD rotation, //for drawing
1965  inT32 block_edge, //edge of block
1966  BOOL8 testing_on //correct orientation
1967  ) {
1968 #ifndef GRAPHICS_DISABLED
1969  ScrollView::Color colour; //of row
1970 #endif
1971  TO_ROW_IT row_it = block->get_rows ();
1972 
1973  row_it.move_to_first ();
1974  for (row_it.mark_cycle_pt (); !row_it.cycled_list (); row_it.forward ()) {
1975  if (row_it.data ()->blob_list ()->empty ())
1976  delete row_it.extract (); //nothing in it
1977  else
1978  fit_parallel_lms (gradient, row_it.data ());
1979  }
1980 #ifndef GRAPHICS_DISABLED
1981  if (testing_on) {
1982  colour = ScrollView::RED;
1983  for (row_it.mark_cycle_pt (); !row_it.cycled_list (); row_it.forward ()) {
1984  plot_parallel_row (row_it.data (), gradient,
1985  block_edge, colour, rotation);
1986  colour = (ScrollView::Color) (colour + 1);
1987  if (colour > ScrollView::MAGENTA)
1988  colour = ScrollView::RED;
1989  }
1990  }
1991 #endif
1992  row_it.sort (row_y_order); //may have gone out of order
1993 }
1994 
1995 
2003 void fit_parallel_lms(float gradient, TO_ROW *row) {
2004  float c; // fitted line
2005  int blobcount; // no of blobs
2006  tesseract::DetLineFit lms;
2007  BLOBNBOX_IT blob_it = row->blob_list();
2008 
2009  blobcount = 0;
2010  for (blob_it.mark_cycle_pt(); !blob_it.cycled_list(); blob_it.forward()) {
2011  if (!blob_it.data()->joined_to_prev()) {
2012  const TBOX& box = blob_it.data()->bounding_box();
2013  lms.Add(ICOORD((box.left() + box.right()) / 2, box.bottom()));
2014  blobcount++;
2015  }
2016  }
2017  double error = lms.ConstrainedFit(gradient, &c);
2018  row->set_parallel_line(gradient, c, error);
2019  if (textord_straight_baselines && blobcount > textord_lms_line_trials) {
2020  error = lms.Fit(&gradient, &c);
2021  }
2022  //set the other too
2023  row->set_line(gradient, c, error);
2024 }
2025 
2026 
2032 namespace tesseract {
2033 void Textord::make_spline_rows(TO_BLOCK *block, // block to do
2034  float gradient, // gradient to fit
2035  BOOL8 testing_on) {
2036 #ifndef GRAPHICS_DISABLED
2037  ScrollView::Color colour; //of row
2038 #endif
2039  TO_ROW_IT row_it = block->get_rows ();
2040 
2041  row_it.move_to_first ();
2042  for (row_it.mark_cycle_pt (); !row_it.cycled_list (); row_it.forward ()) {
2043  if (row_it.data ()->blob_list ()->empty ())
2044  delete row_it.extract (); //nothing in it
2045  else
2046  make_baseline_spline (row_it.data (), block);
2047  }
2048  if (textord_old_baselines) {
2049 #ifndef GRAPHICS_DISABLED
2050  if (testing_on) {
2051  colour = ScrollView::RED;
2052  for (row_it.mark_cycle_pt (); !row_it.cycled_list ();
2053  row_it.forward ()) {
2054  row_it.data ()->baseline.plot (to_win, colour);
2055  colour = (ScrollView::Color) (colour + 1);
2056  if (colour > ScrollView::MAGENTA)
2057  colour = ScrollView::RED;
2058  }
2059  }
2060 #endif
2061  make_old_baselines(block, testing_on, gradient);
2062  }
2063 #ifndef GRAPHICS_DISABLED
2064  if (testing_on) {
2065  colour = ScrollView::RED;
2066  for (row_it.mark_cycle_pt (); !row_it.cycled_list (); row_it.forward ()) {
2067  row_it.data ()->baseline.plot (to_win, colour);
2068  colour = (ScrollView::Color) (colour + 1);
2069  if (colour > ScrollView::MAGENTA)
2070  colour = ScrollView::RED;
2071  }
2072  }
2073 #endif
2074 }
2075 
2076 } // namespace tesseract.
2077 
2078 
2086 void make_baseline_spline(TO_ROW *row, //row to fit
2087  TO_BLOCK *block) {
2088  inT32 *xstarts; // spline boundaries
2089  double *coeffs; // quadratic coeffs
2090  inT32 segments; // no of segments
2091 
2092  xstarts =
2093  (inT32 *) alloc_mem((row->blob_list()->length() + 1) * sizeof(inT32));
2094  if (segment_baseline(row, block, segments, xstarts)
2095  && !textord_straight_baselines && !textord_parallel_baselines) {
2096  coeffs = linear_spline_baseline(row, block, segments, xstarts);
2097  } else {
2098  xstarts[1] = xstarts[segments];
2099  segments = 1;
2100  coeffs = (double *) alloc_mem (3 * sizeof (double));
2101  coeffs[0] = 0;
2102  coeffs[1] = row->line_m ();
2103  coeffs[2] = row->line_c ();
2104  }
2105  row->baseline = QSPLINE (segments, xstarts, coeffs);
2106  free_mem(coeffs);
2107  free_mem(xstarts);
2108 }
2109 
2110 
2118 BOOL8
2119 segment_baseline ( //split baseline
2120 TO_ROW * row, //row to fit
2121 TO_BLOCK * block, //block it came from
2122 inT32 & segments, //no fo segments
2123 inT32 xstarts[] //coords of segments
2124 ) {
2125  BOOL8 needs_curve; //needs curved line
2126  int blobcount; //no of blobs
2127  int blobindex; //current blob
2128  int last_state; //above, on , below
2129  int state; //of current blob
2130  float yshift; //from baseline
2131  TBOX box; //blob box
2132  TBOX new_box; //new_it box
2133  float middle; //xcentre of blob
2134  //blobs
2135  BLOBNBOX_IT blob_it = row->blob_list ();
2136  BLOBNBOX_IT new_it = blob_it; //front end
2137  SORTED_FLOATS yshifts; //shifts from baseline
2138 
2139  needs_curve = FALSE;
2140  box = box_next_pre_chopped (&blob_it);
2141  xstarts[0] = box.left ();
2142  segments = 1;
2143  blobcount = row->blob_list ()->length ();
2144  if (textord_oldbl_debug)
2145  tprintf ("Segmenting baseline of %d blobs at (%d,%d)\n",
2146  blobcount, box.left (), box.bottom ());
2147  if (blobcount <= textord_spline_medianwin
2148  || blobcount < textord_spline_minblobs) {
2149  blob_it.move_to_last ();
2150  box = blob_it.data ()->bounding_box ();
2151  xstarts[1] = box.right ();
2152  return FALSE;
2153  }
2154  last_state = 0;
2155  new_it.mark_cycle_pt ();
2156  for (blobindex = 0; blobindex < textord_spline_medianwin; blobindex++) {
2157  new_box = box_next_pre_chopped (&new_it);
2158  middle = (new_box.left () + new_box.right ()) / 2.0;
2159  yshift = new_box.bottom () - row->line_m () * middle - row->line_c ();
2160  //record shift
2161  yshifts.add (yshift, blobindex);
2162  if (new_it.cycled_list ()) {
2163  xstarts[1] = new_box.right ();
2164  return FALSE;
2165  }
2166  }
2167  for (blobcount = 0; blobcount < textord_spline_medianwin / 2; blobcount++)
2168  box = box_next_pre_chopped (&blob_it);
2169  do {
2170  new_box = box_next_pre_chopped (&new_it);
2171  //get middle one
2172  yshift = yshifts[textord_spline_medianwin / 2];
2173  if (yshift > textord_spline_shift_fraction * block->line_size)
2174  state = 1;
2175  else if (-yshift > textord_spline_shift_fraction * block->line_size)
2176  state = -1;
2177  else
2178  state = 0;
2179  if (state != 0)
2180  needs_curve = TRUE;
2181  // tprintf("State=%d, prev=%d, shift=%g\n",
2182  // state,last_state,yshift);
2183  if (state != last_state && blobcount > textord_spline_minblobs) {
2184  xstarts[segments++] = box.left ();
2185  blobcount = 0;
2186  }
2187  last_state = state;
2188  yshifts.remove (blobindex - textord_spline_medianwin);
2189  box = box_next_pre_chopped (&blob_it);
2190  middle = (new_box.left () + new_box.right ()) / 2.0;
2191  yshift = new_box.bottom () - row->line_m () * middle - row->line_c ();
2192  yshifts.add (yshift, blobindex);
2193  blobindex++;
2194  blobcount++;
2195  }
2196  while (!new_it.cycled_list ());
2197  if (blobcount > textord_spline_minblobs || segments == 1) {
2198  xstarts[segments] = new_box.right ();
2199  }
2200  else {
2201  xstarts[--segments] = new_box.right ();
2202  }
2203  if (textord_oldbl_debug)
2204  tprintf ("Made %d segments on row at (%d,%d)\n",
2205  segments, box.right (), box.bottom ());
2206  return needs_curve;
2207 }
2208 
2209 
2217 double *
2218 linear_spline_baseline ( //split baseline
2219 TO_ROW * row, //row to fit
2220 TO_BLOCK * block, //block it came from
2221 inT32 & segments, //no fo segments
2222 inT32 xstarts[] //coords of segments
2223 ) {
2224  int blobcount; //no of blobs
2225  int blobindex; //current blob
2226  int index1, index2; //blob numbers
2227  int blobs_per_segment; //blobs in each
2228  TBOX box; //blob box
2229  TBOX new_box; //new_it box
2230  //blobs
2231  BLOBNBOX_IT blob_it = row->blob_list ();
2232  BLOBNBOX_IT new_it = blob_it; //front end
2233  float b, c; //fitted curve
2234  tesseract::DetLineFit lms;
2235  double *coeffs; //quadratic coeffs
2236  inT32 segment; //current segment
2237 
2238  box = box_next_pre_chopped (&blob_it);
2239  xstarts[0] = box.left ();
2240  blobcount = 1;
2241  while (!blob_it.at_first ()) {
2242  blobcount++;
2243  box = box_next_pre_chopped (&blob_it);
2244  }
2245  segments = blobcount / textord_spline_medianwin;
2246  if (segments < 1)
2247  segments = 1;
2248  blobs_per_segment = blobcount / segments;
2249  coeffs = (double *) alloc_mem (segments * 3 * sizeof (double));
2250  if (textord_oldbl_debug)
2251  tprintf
2252  ("Linear splining baseline of %d blobs at (%d,%d), into %d segments of %d blobs\n",
2253  blobcount, box.left (), box.bottom (), segments, blobs_per_segment);
2254  segment = 1;
2255  for (index2 = 0; index2 < blobs_per_segment / 2; index2++)
2256  box_next_pre_chopped(&new_it);
2257  index1 = 0;
2258  blobindex = index2;
2259  do {
2260  blobindex += blobs_per_segment;
2261  lms.Clear();
2262  while (index1 < blobindex || (segment == segments && index1 < blobcount)) {
2263  box = box_next_pre_chopped (&blob_it);
2264  int middle = (box.left() + box.right()) / 2;
2265  lms.Add(ICOORD(middle, box.bottom()));
2266  index1++;
2267  if (index1 == blobindex - blobs_per_segment / 2
2268  || index1 == blobcount - 1) {
2269  xstarts[segment] = box.left ();
2270  }
2271  }
2272  lms.Fit(&b, &c);
2273  coeffs[segment * 3 - 3] = 0;
2274  coeffs[segment * 3 - 2] = b;
2275  coeffs[segment * 3 - 1] = c;
2276  segment++;
2277  if (segment > segments)
2278  break;
2279 
2280  blobindex += blobs_per_segment;
2281  lms.Clear();
2282  while (index2 < blobindex || (segment == segments && index2 < blobcount)) {
2283  new_box = box_next_pre_chopped (&new_it);
2284  int middle = (new_box.left() + new_box.right()) / 2;
2285  lms.Add(ICOORD (middle, new_box.bottom()));
2286  index2++;
2287  if (index2 == blobindex - blobs_per_segment / 2
2288  || index2 == blobcount - 1) {
2289  xstarts[segment] = new_box.left ();
2290  }
2291  }
2292  lms.Fit(&b, &c);
2293  coeffs[segment * 3 - 3] = 0;
2294  coeffs[segment * 3 - 2] = b;
2295  coeffs[segment * 3 - 1] = c;
2296  segment++;
2297  }
2298  while (segment <= segments);
2299  return coeffs;
2300 }
2301 
2302 
2309 void assign_blobs_to_rows( //find lines
2310  TO_BLOCK *block, //block to do
2311  float *gradient, //block skew
2312  int pass, //identification
2313  BOOL8 reject_misses, //chuck big ones out
2314  BOOL8 make_new_rows, //add rows for unmatched
2315  BOOL8 drawing_skew //draw smoothed skew
2316  ) {
2317  OVERLAP_STATE overlap_result; //what to do with it
2318  float ycoord; //current y
2319  float top, bottom; //of blob
2320  float g_length = 1.0f; //from gradient
2321  inT16 row_count; //no of rows
2322  inT16 left_x; //left edge
2323  inT16 last_x; //previous edge
2324  float block_skew; //y delta
2325  float smooth_factor; //for new coords
2326  float near_dist; //dist to nearest row
2327  ICOORD testpt; //testing only
2328  BLOBNBOX *blob; //current blob
2329  TO_ROW *row; //current row
2330  TO_ROW *dest_row = NULL; //row to put blob in
2331  //iterators
2332  BLOBNBOX_IT blob_it = &block->blobs;
2333  TO_ROW_IT row_it = block->get_rows ();
2334 
2335  ycoord =
2336  (block->block->bounding_box ().bottom () +
2337  block->block->bounding_box ().top ()) / 2.0f;
2338  if (gradient != NULL)
2339  g_length = sqrt (1 + *gradient * *gradient);
2340 #ifndef GRAPHICS_DISABLED
2341  if (drawing_skew)
2342  to_win->SetCursor(block->block->bounding_box ().left (), ycoord);
2343 #endif
2344  testpt = ICOORD (textord_test_x, textord_test_y);
2345  blob_it.sort (blob_x_order);
2346  smooth_factor = 1.0;
2347  block_skew = 0.0f;
2348  row_count = row_it.length (); //might have rows
2349  if (!blob_it.empty ()) {
2350  left_x = blob_it.data ()->bounding_box ().left ();
2351  }
2352  else {
2353  left_x = block->block->bounding_box ().left ();
2354  }
2355  last_x = left_x;
2356  for (blob_it.mark_cycle_pt (); !blob_it.cycled_list (); blob_it.forward ()) {
2357  blob = blob_it.data ();
2358  if (gradient != NULL) {
2359  block_skew = (1 - 1 / g_length) * blob->bounding_box ().bottom ()
2360  + *gradient / g_length * blob->bounding_box ().left ();
2361  }
2362  else if (blob->bounding_box ().left () - last_x > block->line_size / 2
2363  && last_x - left_x > block->line_size * 2
2364  && textord_interpolating_skew) {
2365  // tprintf("Interpolating skew from %g",block_skew);
2366  block_skew *= (float) (blob->bounding_box ().left () - left_x)
2367  / (last_x - left_x);
2368  // tprintf("to %g\n",block_skew);
2369  }
2370  last_x = blob->bounding_box ().left ();
2371  top = blob->bounding_box ().top () - block_skew;
2372  bottom = blob->bounding_box ().bottom () - block_skew;
2373 #ifndef GRAPHICS_DISABLED
2374  if (drawing_skew)
2375  to_win->DrawTo(blob->bounding_box ().left (), ycoord + block_skew);
2376 #endif
2377  if (!row_it.empty ()) {
2378  for (row_it.move_to_first ();
2379  !row_it.at_last () && row_it.data ()->min_y () > top;
2380  row_it.forward ());
2381  row = row_it.data ();
2382  if (row->min_y () <= top && row->max_y () >= bottom) {
2383  //any overlap
2384  dest_row = row;
2385  overlap_result = most_overlapping_row (&row_it, dest_row,
2386  top, bottom,
2387  block->line_size,
2388  blob->bounding_box ().
2389  contains (testpt));
2390  if (overlap_result == NEW_ROW && !reject_misses)
2391  overlap_result = ASSIGN;
2392  }
2393  else {
2394  overlap_result = NEW_ROW;
2395  if (!make_new_rows) {
2396  near_dist = row_it.data_relative (-1)->min_y () - top;
2397  //below bottom
2398  if (bottom < row->min_y ()) {
2399  if (row->min_y () - bottom <=
2400  (block->line_spacing -
2401  block->line_size) * tesseract::CCStruct::kDescenderFraction) {
2402  //done it
2403  overlap_result = ASSIGN;
2404  dest_row = row;
2405  }
2406  }
2407  else if (near_dist > 0
2408  && near_dist < bottom - row->max_y ()) {
2409  row_it.backward ();
2410  dest_row = row_it.data ();
2411  if (dest_row->min_y () - bottom <=
2412  (block->line_spacing -
2413  block->line_size) * tesseract::CCStruct::kDescenderFraction) {
2414  //done it
2415  overlap_result = ASSIGN;
2416  }
2417  }
2418  else {
2419  if (top - row->max_y () <=
2420  (block->line_spacing -
2421  block->line_size) * (textord_overlap_x +
2422  tesseract::CCStruct::kAscenderFraction)) {
2423  //done it
2424  overlap_result = ASSIGN;
2425  dest_row = row;
2426  }
2427  }
2428  }
2429  }
2430  if (overlap_result == ASSIGN)
2431  dest_row->add_blob (blob_it.extract (), top, bottom,
2432  block->line_size);
2433  if (overlap_result == NEW_ROW) {
2434  if (make_new_rows && top - bottom < block->max_blob_size) {
2435  dest_row =
2436  new TO_ROW (blob_it.extract (), top, bottom,
2437  block->line_size);
2438  row_count++;
2439  if (bottom > row_it.data ()->min_y ())
2440  row_it.add_before_then_move (dest_row);
2441  //insert in right place
2442  else
2443  row_it.add_after_then_move (dest_row);
2444  smooth_factor =
2445  1.0 / (row_count * textord_skew_lag +
2446  textord_skewsmooth_offset);
2447  }
2448  else
2449  overlap_result = REJECT;
2450  }
2451  }
2452  else if (make_new_rows && top - bottom < block->max_blob_size) {
2453  overlap_result = NEW_ROW;
2454  dest_row =
2455  new TO_ROW(blob_it.extract(), top, bottom, block->line_size);
2456  row_count++;
2457  row_it.add_after_then_move(dest_row);
2458  smooth_factor = 1.0 / (row_count * textord_skew_lag +
2459  textord_skewsmooth_offset2);
2460  }
2461  else
2462  overlap_result = REJECT;
2463  if (blob->bounding_box ().contains(testpt) && textord_debug_blob) {
2464  if (overlap_result != REJECT) {
2465  tprintf("Test blob assigned to row at (%g,%g) on pass %d\n",
2466  dest_row->min_y(), dest_row->max_y(), pass);
2467  }
2468  else {
2469  tprintf("Test blob assigned to no row on pass %d\n", pass);
2470  }
2471  }
2472  if (overlap_result != REJECT) {
2473  while (!row_it.at_first() &&
2474  row_it.data()->min_y() > row_it.data_relative(-1)->min_y()) {
2475  row = row_it.extract();
2476  row_it.backward();
2477  row_it.add_before_then_move(row);
2478  }
2479  while (!row_it.at_last() &&
2480  row_it.data ()->min_y() < row_it.data_relative (1)->min_y()) {
2481  row = row_it.extract();
2482  row_it.forward();
2483  // Keep rows in order.
2484  row_it.add_after_then_move(row);
2485  }
2486  BLOBNBOX_IT added_blob_it(dest_row->blob_list());
2487  added_blob_it.move_to_last();
2488  TBOX prev_box = added_blob_it.data_relative(-1)->bounding_box();
2489  if (dest_row->blob_list()->singleton() ||
2490  !prev_box.major_x_overlap(blob->bounding_box())) {
2491  block_skew = (1 - smooth_factor) * block_skew
2492  + smooth_factor * (blob->bounding_box().bottom() -
2493  dest_row->initial_min_y());
2494  }
2495  }
2496  }
2497  for (row_it.mark_cycle_pt(); !row_it.cycled_list(); row_it.forward()) {
2498  if (row_it.data()->blob_list()->empty())
2499  delete row_it.extract(); // Discard empty rows.
2500  }
2501 }
2502 
2503 
2509 OVERLAP_STATE most_overlapping_row( //find best row
2510  TO_ROW_IT *row_it, //iterator
2511  TO_ROW *&best_row, //output row
2512  float top, //top of blob
2513  float bottom, //bottom of blob
2514  float rowsize, //max row size
2515  BOOL8 testing_blob //test stuff
2516  ) {
2517  OVERLAP_STATE result; //result of tests
2518  float overlap; //of blob & row
2519  float bestover; //nearest row
2520  float merge_top, merge_bottom; //size of merged row
2521  ICOORD testpt; //testing only
2522  TO_ROW *row; //current row
2523  TO_ROW *test_row; //for multiple overlaps
2524  BLOBNBOX_IT blob_it; //for merging rows
2525 
2526  result = ASSIGN;
2527  row = row_it->data ();
2528  bestover = top - bottom;
2529  if (top > row->max_y ())
2530  bestover -= top - row->max_y ();
2531  if (bottom < row->min_y ())
2532  //compute overlap
2533  bestover -= row->min_y () - bottom;
2534  if (testing_blob && textord_debug_blob) {
2535  tprintf("Test blob y=(%g,%g), row=(%f,%f), size=%g, overlap=%f\n",
2536  bottom, top, row->min_y(), row->max_y(), rowsize, bestover);
2537  }
2538  test_row = row;
2539  do {
2540  if (!row_it->at_last ()) {
2541  row_it->forward ();
2542  test_row = row_it->data ();
2543  if (test_row->min_y () <= top && test_row->max_y () >= bottom) {
2544  merge_top =
2545  test_row->max_y () >
2546  row->max_y ()? test_row->max_y () : row->max_y ();
2547  merge_bottom =
2548  test_row->min_y () <
2549  row->min_y ()? test_row->min_y () : row->min_y ();
2550  if (merge_top - merge_bottom <= rowsize) {
2551  if (testing_blob) {
2552  tprintf ("Merging rows at (%g,%g), (%g,%g)\n",
2553  row->min_y (), row->max_y (),
2554  test_row->min_y (), test_row->max_y ());
2555  }
2556  test_row->set_limits (merge_bottom, merge_top);
2557  blob_it.set_to_list (test_row->blob_list ());
2558  blob_it.add_list_after (row->blob_list ());
2559  blob_it.sort (blob_x_order);
2560  row_it->backward ();
2561  delete row_it->extract ();
2562  row_it->forward ();
2563  bestover = -1.0f; //force replacement
2564  }
2565  overlap = top - bottom;
2566  if (top > test_row->max_y ())
2567  overlap -= top - test_row->max_y ();
2568  if (bottom < test_row->min_y ())
2569  overlap -= test_row->min_y () - bottom;
2570  if (bestover >= rowsize - 1 && overlap >= rowsize - 1) {
2571  result = REJECT;
2572  }
2573  if (overlap > bestover) {
2574  bestover = overlap; //find biggest overlap
2575  row = test_row;
2576  }
2577  if (testing_blob && textord_debug_blob) {
2578  tprintf("Test blob y=(%g,%g), row=(%f,%f), size=%g, overlap=%f->%f\n",
2579  bottom, top, test_row->min_y(), test_row->max_y(),
2580  rowsize, overlap, bestover);
2581  }
2582  }
2583  }
2584  }
2585  while (!row_it->at_last ()
2586  && test_row->min_y () <= top && test_row->max_y () >= bottom);
2587  while (row_it->data () != row)
2588  row_it->backward (); //make it point to row
2589  //doesn't overlap much
2590  if (top - bottom - bestover > rowsize * textord_overlap_x &&
2591  (!textord_fix_makerow_bug || bestover < rowsize * textord_overlap_x)
2592  && result == ASSIGN)
2593  result = NEW_ROW; //doesn't overlap enough
2594  best_row = row;
2595  return result;
2596 }
2597 
2598 
2604 int blob_x_order( //sort function
2605  const void *item1, //items to compare
2606  const void *item2) {
2607  //converted ptr
2608  BLOBNBOX *blob1 = *(BLOBNBOX **) item1;
2609  //converted ptr
2610  BLOBNBOX *blob2 = *(BLOBNBOX **) item2;
2611 
2612  if (blob1->bounding_box ().left () < blob2->bounding_box ().left ())
2613  return -1;
2614  else if (blob1->bounding_box ().left () > blob2->bounding_box ().left ())
2615  return 1;
2616  else
2617  return 0;
2618 }
2619 
2620 
2626 int row_y_order( //sort function
2627  const void *item1, //items to compare
2628  const void *item2) {
2629  //converted ptr
2630  TO_ROW *row1 = *(TO_ROW **) item1;
2631  //converted ptr
2632  TO_ROW *row2 = *(TO_ROW **) item2;
2633 
2634  if (row1->parallel_c () > row2->parallel_c ())
2635  return -1;
2636  else if (row1->parallel_c () < row2->parallel_c ())
2637  return 1;
2638  else
2639  return 0;
2640 }
2641 
2642 
2648 int row_spacing_order( //sort function
2649  const void *item1, //items to compare
2650  const void *item2) {
2651  //converted ptr
2652  TO_ROW *row1 = *(TO_ROW **) item1;
2653  //converted ptr
2654  TO_ROW *row2 = *(TO_ROW **) item2;
2655 
2656  if (row1->spacing < row2->spacing)
2657  return -1;
2658  else if (row1->spacing > row2->spacing)
2659  return 1;
2660  else
2661  return 0;
2662 }
2663 
2670 void mark_repeated_chars(TO_ROW *row) {
2671  BLOBNBOX_IT box_it(row->blob_list()); // Iterator.
2672  int num_repeated_sets = 0;
2673  if (!box_it.empty()) {
2674  do {
2675  BLOBNBOX* bblob = box_it.data();
2676  int repeat_length = 1;
2677  if (bblob->flow() == BTFT_LEADER &&
2678  !bblob->joined_to_prev() && bblob->cblob() != NULL) {
2679  BLOBNBOX_IT test_it(box_it);
2680  for (test_it.forward(); !test_it.at_first();) {
2681  bblob = test_it.data();
2682  if (bblob->flow() != BTFT_LEADER)
2683  break;
2684  test_it.forward();
2685  bblob = test_it.data();
2686  if (bblob->joined_to_prev() || bblob->cblob() == NULL) {
2687  repeat_length = 0;
2688  break;
2689  }
2690  ++repeat_length;
2691  }
2692  }
2693  if (repeat_length >= kMinLeaderCount) {
2694  num_repeated_sets++;
2695  for (; repeat_length > 0; box_it.forward(), --repeat_length) {
2696  bblob = box_it.data();
2697  bblob->set_repeated_set(num_repeated_sets);
2698  }
2699  } else {
2700  bblob->set_repeated_set(0);
2701  box_it.forward();
2702  }
2703  } while (!box_it.at_first()); // until all done
2704  }
2705  row->set_num_repeated_sets(num_repeated_sets);
2706 }
textord_show_expanded_rows
bool textord_show_expanded_rows
Definition: makerow.cpp:47
BLOBNBOX::set_repeated_set
void set_repeated_set(int set_id)
Definition: blobbox.h:250
BLOBNBOX::merge
void merge(BLOBNBOX *nextblob)
Definition: blobbox.cpp:87
BLOBNBOX::joined_to_prev
bool joined_to_prev() const
Definition: blobbox.h:241
textord_heavy_nr
bool textord_heavy_nr
Definition: makerow.cpp:44
deskew_block_coords
TBOX deskew_block_coords(TO_BLOCK *block, float gradient)
Definition: makerow.cpp:745
expand_rows
void expand_rows(ICOORD page_tr, TO_BLOCK *block, float gradient, FCOORD rotation, inT32 block_edge, BOOL8 testing_on)
Definition: makerow.cpp:975
textord_descheight_mode_fraction
double textord_descheight_mode_fraction
Definition: makerow.cpp:95
textord_debug_blob
bool textord_debug_blob
Definition: makerow.cpp:103
textord_test_landscape
bool textord_test_landscape
Definition: makerow.cpp:50
TRUE
#define TRUE
Definition: capi.h:45
TO_ROW::blob_list
BLOBNBOX_LIST * blob_list()
Definition: blobbox.h:595
TBOX::rotate
void rotate(const FCOORD &vec)
Definition: rect.h:189
textord_occupancy_threshold
double textord_occupancy_threshold
Definition: makerow.cpp:86
mark_repeated_chars
void mark_repeated_chars(TO_ROW *row)
Definition: makerow.cpp:2670
TBOX::bottom
inT16 bottom() const
Definition: rect.h:61
textord_spline_medianwin
int textord_spline_medianwin
Definition: makerow.cpp:66
inT8
SIGNED char inT8
Definition: host.h:98
MAX
#define MAX(x, y)
Definition: ndminx.h:24
compute_dropout_distances
void compute_dropout_distances(inT32 *occupation, inT32 *thresholds, inT32 line_count)
Definition: makerow.cpp:928
textord_old_baselines
bool textord_old_baselines
Definition: makerow.cpp:53
textord_fix_xheight_bug
bool textord_fix_xheight_bug
Definition: makerow.cpp:55
STATS::mode
inT32 mode() const
Definition: statistc.cpp:118
TO_ROW::set_num_repeated_sets
void set_num_repeated_sets(int num_sets)
Definition: blobbox.h:636
TO_ROW::set_line
void set_line(float new_m, float new_c, float new_error)
Definition: blobbox.h:599
textord_show_parallel_rows
bool textord_show_parallel_rows
Definition: makerow.cpp:46
kMinLeaderCount
const int kMinLeaderCount
Definition: makerow.cpp:107
make_initial_textrows
void make_initial_textrows(ICOORD page_tr, TO_BLOCK *block, FCOORD rotation, BOOL8 testing_on)
Definition: makerow.cpp:227
STATS::max_bucket
inT32 max_bucket() const
Definition: statistc.cpp:224
compute_xheight_from_modes
int compute_xheight_from_modes(STATS *heights, STATS *floating_heights, bool cap_only, int min_height, int max_height, float *xheight, float *ascrise)
Definition: makerow.cpp:1497
OVERLAP_STATE
OVERLAP_STATE
Definition: makerow.h:29
MAX_HEIGHT_MODES
#define MAX_HEIGHT_MODES
Definition: makerow.cpp:105
MAX_INT32
#define MAX_INT32
Definition: host.h:120
adjust_row_limits
void adjust_row_limits(TO_BLOCK *block)
Definition: makerow.cpp:1133
TO_BLOCK::line_spacing
float line_spacing
Definition: blobbox.h:775
TO_ROW::min_y
float min_y() const
Definition: blobbox.h:557
textord_min_xheight
int textord_min_xheight
Definition: makerow.cpp:69
textord_fix_makerow_bug
bool textord_fix_makerow_bug
Definition: makerow.cpp:56
choose_nth_item
inT32 choose_nth_item(inT32 index, float *array, inT32 count)
Definition: statistc.cpp:641
textord_show_final_blobs
bool textord_show_final_blobs
Definition: makerow.cpp:49
TO_BLOCK::underlines
BLOBNBOX_LIST underlines
Definition: blobbox.h:769
double_VAR
#define double_VAR(name, val, comment)
Definition: params.h:286
compute_line_occupation
void compute_line_occupation(TO_BLOCK *block, float gradient, inT32 min_y, inT32 max_y, inT32 *occupation, inT32 *deltas)
Definition: makerow.cpp:781
blob_x_order
int blob_x_order(const void *item1, const void *item2)
Definition: makerow.cpp:2604
TO_ROW::set_limits
void set_limits(float new_min, float new_max)
Definition: blobbox.h:618
row_spacing_order
int row_spacing_order(const void *item1, const void *item2)
Definition: makerow.cpp:2648
TO_ROW::max_y
float max_y() const
Definition: blobbox.h:554
textord_xheight_mode_fraction
double textord_xheight_mode_fraction
Definition: makerow.cpp:91
inT32
int inT32
Definition: host.h:102
segment_baseline
BOOL8 segment_baseline(TO_ROW *row, TO_BLOCK *block, inT32 &segments, inT32 xstarts[])
Definition: makerow.cpp:2119
FCOORD
Definition: points.h:189
C_OUTLINE::start_pos
const ICOORD & start_pos() const
Definition: coutln.h:146
ScrollView::SetCursor
void SetCursor(int x, int y)
Definition: scrollview.cpp:525
C_BLOB::CheckInverseFlagAndDirection
void CheckInverseFlagAndDirection()
Definition: stepblob.cpp:221
compute_page_skew
void compute_page_skew(TO_BLOCK_LIST *blocks, float &page_m, float &page_err)
Definition: makerow.cpp:287
TO_BLOCK::xheight
float xheight
Definition: blobbox.h:784
STATS::get_total
inT32 get_total() const
Definition: statistc.h:86
kMinSize
const int kMinSize
Definition: makerow.cpp:384
TO_ROW::believability
float believability() const
Definition: blobbox.h:581
TO_BLOCK::large_blobs
BLOBNBOX_LIST large_blobs
Definition: blobbox.h:772
test_underline
BOOL8 test_underline(BOOL8 testing_on, C_BLOB *blob, inT16 baseline, inT16 xheight)
Definition: blkocc.cpp:53
C_OUTLINE::deep_copy
static C_OUTLINE * deep_copy(const C_OUTLINE *src)
Definition: coutln.h:259
tesseract::DetLineFit::Fit
double Fit(ICOORD *pt1, ICOORD *pt2)
Definition: detlinefit.h:75
get_min_max_xheight
void get_min_max_xheight(int block_linesize, int *min_height, int *max_height)
Definition: makerow.h:116
TBOX::height
inT16 height() const
Definition: rect.h:104
separate_underlines
void separate_underlines(TO_BLOCK *block, float gradient, FCOORD rotation, BOOL8 testing_on)
Definition: makerow.cpp:1802
C_OUTLINE::step
ICOORD step(int index) const
Definition: coutln.h:142
FLOAT32
float FLOAT32
Definition: host.h:111
TO_BLOCK::max_blob_size
float max_blob_size
Definition: blobbox.h:782
textord_skew_ile
double textord_skew_ile
Definition: makerow.cpp:74
textord_min_linesize
double textord_min_linesize
Definition: makerow.cpp:83
textord_debug_xheights
bool textord_debug_xheights
Definition: makerow.cpp:57
textord_min_blobs_in_row
int textord_min_blobs_in_row
Definition: makerow.cpp:64
textord_interpolating_skew
bool textord_interpolating_skew
Definition: makerow.cpp:59
tesseract::CCStruct::kAscenderFraction
static const double kAscenderFraction
Definition: ccstruct.h:35
textord_max_blob_overlaps
int textord_max_blob_overlaps
Definition: makerow.cpp:68
textord_descx_ratio_min
double textord_descx_ratio_min
Definition: makerow.cpp:98
tprintf
#define tprintf(...)
Definition: tprintf.h:31
fit_lms_line
void fit_lms_line(TO_ROW *row)
Definition: makerow.cpp:267
ScrollView::Pen
void Pen(Color color)
Definition: scrollview.cpp:726
get_row_category
ROW_CATEGORY get_row_category(const TO_ROW *row)
Definition: makerow.h:123
TO_BLOCK::blobs
BLOBNBOX_LIST blobs
Definition: blobbox.h:768
FCOORD::rotate
void rotate(const FCOORD vec)
Definition: ipoints.h:471
make_single_row
float make_single_row(ICOORD page_tr, bool allow_sub_blobs, TO_BLOCK *block, TO_BLOCK_LIST *blocks)
Definition: makerow.cpp:164
textord_spline_outlier_fraction
double textord_spline_outlier_fraction
Definition: makerow.cpp:73
TO_ROW::parallel_c
float parallel_c() const
Definition: blobbox.h:575
textord_new_initial_xheight
bool textord_new_initial_xheight
Definition: makerow.cpp:102
TO_ROW::xheight
float xheight
Definition: blobbox.h:653
INT_VAR
#define INT_VAR(name, val, comment)
Definition: params.h:277
box_next_pre_chopped
TBOX box_next_pre_chopped(BLOBNBOX_IT *it)
Definition: blobbox.cpp:660
TBOX
Definition: rect.h:30
BLOBNBOX::bounding_box
const TBOX & bounding_box() const
Definition: blobbox.h:215
tesseract::CCStruct::kDescenderFraction
static const double kDescenderFraction
Definition: ccstruct.h:33
TO_BLOCK::key_row
TO_ROW * key_row
Definition: blobbox.h:794
BLOBNBOX::repeated_set
int repeated_set() const
Definition: blobbox.h:247
STATS::min_bucket
inT32 min_bucket() const
Definition: statistc.cpp:209
textord_underline_width
double textord_underline_width
Definition: makerow.cpp:87
FALSE
#define FALSE
Definition: capi.h:46
linear_spline_baseline
double * linear_spline_baseline(TO_ROW *row, TO_BLOCK *block, inT32 &segments, inT32 xstarts[])
Definition: makerow.cpp:2218
textord_excess_blobsize
double textord_excess_blobsize
Definition: makerow.cpp:85
textord_expansion_factor
double textord_expansion_factor
Definition: makerow.cpp:80
textord_ascx_ratio_min
double textord_ascx_ratio_min
Definition: makerow.cpp:96
textord_straight_baselines
bool textord_straight_baselines
Definition: makerow.cpp:52
most_overlapping_row
OVERLAP_STATE most_overlapping_row(TO_ROW_IT *row_it, TO_ROW *&best_row, float top, float bottom, float rowsize, BOOL8 testing_blob)
Definition: makerow.cpp:2509
draw_occupation
void draw_occupation(inT32 xleft, inT32 ybottom, inT32 min_y, inT32 max_y, inT32 occupation[], inT32 thresholds[])
Definition: drawtord.cpp:166
PDBLK::bounding_box
void bounding_box(ICOORD &bottom_left, ICOORD &top_right) const
get box
Definition: pdblock.h:59
TBOX::right
inT16 right() const
Definition: rect.h:75
inT16
short inT16
Definition: host.h:100
ROW_CATEGORY
ROW_CATEGORY
Definition: makerow.h:36
STATS
Definition: statistc.h:33
QSPLINE::y
double y(double x) const
Definition: quspline.cpp:217
ScrollView::Rectangle
void Rectangle(int x1, int y1, int x2, int y2)
Definition: scrollview.cpp:606
row_y_order
int row_y_order(const void *item1, const void *item2)
Definition: makerow.cpp:2626
TO_ROW::baseline
QSPLINE baseline
Definition: blobbox.h:666
STATS::pile_count
inT32 pile_count(inT32 value) const
Definition: statistc.h:78
ABORT
Definition: errcode.h:30
textord_skewsmooth_offset2
int textord_skewsmooth_offset2
Definition: makerow.cpp:61
TO_ROW::xheight_evidence
int xheight_evidence
Definition: blobbox.h:654
alloc_mem
void * alloc_mem(inT32 count)
Definition: memry.cpp:47
tesseract::DetLineFit::ConstrainedFit
double ConstrainedFit(const FCOORD &direction, double min_dist, double max_dist, bool debug, ICOORD *line_pt)
Definition: detlinefit.cpp:131
textord_test_x
int textord_test_x
Definition: makerow.cpp:62
TO_BLOCK::noise_blobs
BLOBNBOX_LIST noise_blobs
Definition: blobbox.h:770
TO_BLOCK::block
BLOCK * block
Definition: blobbox.h:773
MIN
#define MIN(x, y)
Definition: ndminx.h:28
plot_to_row
void plot_to_row(TO_ROW *row, ScrollView::Color colour, FCOORD rotation)
Definition: drawtord.cpp:91
TBOX::left
inT16 left() const
Definition: rect.h:68
TO_ROW::line_error
float line_error() const
Definition: blobbox.h:572
FCOORD::y
float y() const
Definition: points.h:212
ICOORD::x
inT16 x() const
access function
Definition: points.h:52
textord_xheight_error_margin
double textord_xheight_error_margin
Definition: makerow.cpp:100
textord_lms_line_trials
int textord_lms_line_trials
Definition: makerow.cpp:101
textord_minxh
double textord_minxh
Definition: makerow.cpp:82
compute_occupation_threshold
void compute_occupation_threshold(inT32 low_window, inT32 high_window, inT32 line_count, inT32 *occupation, inT32 *thresholds)
Definition: makerow.cpp:847
textord_spline_minblobs
int textord_spline_minblobs
Definition: makerow.cpp:65
C_BLOB::FakeBlob
static C_BLOB * FakeBlob(const TBOX &box)
Definition: stepblob.cpp:238
C_BLOB::out_list
C_OUTLINE_LIST * out_list()
Definition: stepblob.h:64
MEMORY_OUT
const ERRCODE MEMORY_OUT
Definition: stderr.h:25
SORTED_FLOATS::remove
void remove(inT32 key)
Definition: sortflts.cpp:53
textord_test_y
int textord_test_y
Definition: makerow.cpp:63
tesseract::Textord::compute_row_xheight
void compute_row_xheight(TO_ROW *row, const FCOORD &rotation, float gradient, int block_line_size)
Definition: makerow.cpp:1396
compute_height_modes
inT32 compute_height_modes(STATS *heights, inT32 min_height, inT32 max_height, inT32 *modes, inT32 maxmodes)
Definition: makerow.cpp:1653
TBOX::print
void print() const
Definition: rect.h:270
TO_ROW::descdrop
float descdrop
Definition: blobbox.h:656
INT32FORMAT
#define INT32FORMAT
Definition: host.h:115
ICOORD
integer coordinate
Definition: points.h:30
TO_ROW::line_m
float line_m() const
Definition: blobbox.h:566
textord_skewsmooth_offset
int textord_skewsmooth_offset
Definition: makerow.cpp:60
textord_ascheight_mode_fraction
double textord_ascheight_mode_fraction
Definition: makerow.cpp:93
TO_ROW::num_repeated_sets
int num_repeated_sets() const
Definition: blobbox.h:633
assign_blobs_to_rows
void assign_blobs_to_rows(TO_BLOCK *block, float *gradient, int pass, BOOL8 reject_misses, BOOL8 make_new_rows, BOOL8 drawing_skew)
Definition: makerow.cpp:2309
kNoiseSize
const double kNoiseSize
Definition: makerow.cpp:383
textord_parallel_baselines
bool textord_parallel_baselines
Definition: makerow.cpp:51
find_best_dropout_row
BOOL8 find_best_dropout_row(TO_ROW *row, inT32 distance, float dist_limit, inT32 line_index, TO_ROW_IT *row_it, BOOL8 testing_on)
Definition: makerow.cpp:665
TBOX::major_overlap
bool major_overlap(const TBOX &box) const
Definition: rect.h:358
textord_biased_skewcalc
bool textord_biased_skewcalc
Definition: makerow.cpp:58
tesseract::Textord::textord_single_height_mode
bool textord_single_height_mode
Definition: textord.h:261
TO_BLOCK::small_blobs
BLOBNBOX_LIST small_blobs
Definition: blobbox.h:771
ASSIGN
Definition: makerow.h:31
TBOX::major_x_overlap
bool major_x_overlap(const TBOX &box) const
Definition: rect.h:402
BLOBNBOX::flow
BlobTextFlowType flow() const
Definition: blobbox.h:280
textord_linespace_iqrlimit
double textord_linespace_iqrlimit
Definition: makerow.cpp:76
REJECT
Definition: makerow.h:32
TO_BLOCK::line_size
float line_size
Definition: blobbox.h:781
TO_BLOCK::get_rows
TO_ROW_LIST * get_rows()
Definition: blobbox.h:700
POLY_BLOCK::IsText
bool IsText() const
Definition: polyblk.h:52
make_baseline_spline
void make_baseline_spline(TO_ROW *row, TO_BLOCK *block)
Definition: makerow.cpp:2086
TO_ROW::initial_min_y
float initial_min_y() const
Definition: blobbox.h:563
tesseract::DetLineFit::Add
void Add(const ICOORD &pt)
Definition: detlinefit.cpp:52
TO_ROW::intercept
float intercept() const
Definition: blobbox.h:584
TO_ROW::all_caps
BOOL8 all_caps
Definition: blobbox.h:642
FCOORD::x
float x() const
Definition: points.h:209
ICOORD::y
inT16 y() const
access_function
Definition: points.h:56
TBOX::top
inT16 top() const
Definition: rect.h:54
TO_BLOCK::baseline_offset
float baseline_offset
Definition: blobbox.h:783
fit_parallel_rows
void fit_parallel_rows(TO_BLOCK *block, float gradient, FCOORD rotation, inT32 block_edge, BOOL8 testing_on)
Definition: makerow.cpp:1961
tesseract::Textord::make_spline_rows
void make_spline_rows(TO_BLOCK *block, float gradient, BOOL8 testing_on)
Definition: makerow.cpp:2033
BLOCK::classify_rotation
FCOORD classify_rotation() const
Definition: ocrblock.h:144
textord_old_xheight
bool textord_old_xheight
Definition: makerow.cpp:54
textord_show_final_rows
bool textord_show_final_rows
Definition: makerow.cpp:48
TO_ROW::line_c
float line_c() const
Definition: blobbox.h:569
BOOL8
unsigned char BOOL8
Definition: host.h:113
within_error_margin
bool within_error_margin(float test, float num, float margin)
Definition: makerow.h:129
vigorous_noise_removal
void vigorous_noise_removal(TO_BLOCK *block)
Definition: makerow.cpp:473
tesseract::CCStruct::kXHeightFraction
static const double kXHeightFraction
Definition: ccstruct.h:34
correct_row_xheight
void correct_row_xheight(TO_ROW *row, float xheight, float ascrise, float descdrop)
Definition: makerow.cpp:1715
plot_parallel_row
void plot_parallel_row(TO_ROW *row, float gradient, inT32 left, ScrollView::Color colour, FCOORD rotation)
Definition: drawtord.cpp:125
TBOX::contains
bool contains(const FCOORD pt) const
Definition: rect.h:323
delete_non_dropout_rows
void delete_non_dropout_rows(TO_BLOCK *block, float gradient, FCOORD rotation, inT32 block_edge, BOOL8 testing_on)
Definition: makerow.cpp:577
textord_chop_width
double textord_chop_width
Definition: makerow.cpp:78
pre_associate_blobs
void pre_associate_blobs(ICOORD page_tr, TO_BLOCK *block, FCOORD rotation, BOOL8 testing_on)
Definition: makerow.cpp:1875
crotate_cblob
C_BLOB * crotate_cblob(C_BLOB *blob, FCOORD rotation)
Definition: blobbox.cpp:606
ASSERT_HOST
#define ASSERT_HOST(x)
Definition: errcode.h:84
textord_skew_lag
double textord_skew_lag
Definition: makerow.cpp:75
SORTED_FLOATS::add
void add(float value, inT32 key)
Definition: sortflts.cpp:28
to_win
EXTERN ScrollView * to_win
Definition: drawtord.cpp:38
compute_row_stats
void compute_row_stats(TO_BLOCK *block, BOOL8 testing_on)
Definition: makerow.cpp:1169
textord_show_initial_rows
bool textord_show_initial_rows
Definition: makerow.cpp:45
make_rows
float make_rows(ICOORD page_tr, TO_BLOCK_LIST *port_blocks)
Definition: makerow.cpp:201
textord_min_blob_height_fraction
double textord_min_blob_height_fraction
Definition: makerow.cpp:89
compute_row_descdrop
inT32 compute_row_descdrop(TO_ROW *row, float gradient, int xheight_blob_count, STATS *asc_heights)
Definition: makerow.cpp:1593
textord_width_limit
double textord_width_limit
Definition: makerow.cpp:77
textord_ascx_ratio_max
double textord_ascx_ratio_max
Definition: makerow.cpp:97
TO_ROW::rep_chars_marked
bool rep_chars_marked() const
Definition: blobbox.h:627
BLOBNBOX::chop
void chop(BLOBNBOX_IT *start_it, BLOBNBOX_IT *blob_it, FCOORD rotation, float xheight)
Definition: blobbox.cpp:115
TO_ROW::ascrise
float ascrise
Definition: blobbox.h:655
fill_heights
void fill_heights(TO_ROW *row, float gradient, int min_height, int max_height, STATS *heights, STATS *floating_heights)
Definition: makerow.cpp:1436
BOOL_VAR
#define BOOL_VAR(name, val, comment)
Definition: params.h:280
TBOX::width
inT16 width() const
Definition: rect.h:111
textord_spline_shift_fraction
double textord_spline_shift_fraction
Definition: makerow.cpp:71
TO_ROW::merged
BOOL8 merged
Definition: blobbox.h:641
MakeRowFromSubBlobs
float MakeRowFromSubBlobs(TO_BLOCK *block, C_BLOB *blob, TO_ROW_IT *row_it)
Definition: makerow.cpp:137
TO_ROW::set_parallel_line
void set_parallel_line(float gradient, float new_c, float new_error)
Definition: blobbox.h:607
ERRCODE::error
void error(const char *caller, TessErrorLogCode action, const char *format,...) const
Definition: errcode.cpp:40
create_to_win
ScrollView * create_to_win(ICOORD page_tr)
Definition: drawtord.cpp:47
STATS::median
double median() const
Definition: statistc.cpp:242
ScrollView::DrawTo
void DrawTo(int x, int y)
Definition: scrollview.cpp:531
textord_overlap_x
double textord_overlap_x
Definition: makerow.cpp:81
free_mem
void free_mem(void *oldchunk)
Definition: memry.cpp:55
TO_ROW::spacing
float spacing
Definition: blobbox.h:652
cleanup_rows_making
void cleanup_rows_making(ICOORD page_tr, TO_BLOCK *block, float gradient, FCOORD rotation, inT32 block_edge, BOOL8 testing_on)
Definition: makerow.cpp:524
BLOBNBOX::cblob
C_BLOB * cblob() const
Definition: blobbox.h:253
STATS::add
void add(inT32 value, inT32 count)
Definition: statistc.cpp:104
STATS::print
void print() const
Definition: statistc.cpp:537
TO_ROW::add_blob
void add_blob(BLOBNBOX *blob, float top, float bottom, float row_size)
Definition: blobbox.cpp:728
textord_oldbl_debug
EXTERN bool textord_oldbl_debug
Definition: oldbasel.cpp:39
C_OUTLINE::pathlength
inT32 pathlength() const
Definition: coutln.h:133
tesseract::CCStruct::kXHeightCapRatio
static const double kXHeightCapRatio
Definition: ccstruct.h:37
textord_descx_ratio_max
double textord_descx_ratio_max
Definition: makerow.cpp:99
fit_parallel_lms
void fit_parallel_lms(float gradient, TO_ROW *row)
Definition: makerow.cpp:2003
tesseract::Textord::compute_block_xheight
void compute_block_xheight(TO_BLOCK *block, float gradient)
Definition: makerow.cpp:1284