tospace.cpp

一ＯＣＲ的相关资料。．希望对研究ＯＣＲ的朋友有所帮助．

#include "mfcpch.h"
#include          "tovars.h"
#include          "drawtord.h"
#include          "tospace.h"
#include          "ndminx.h"
#include          "statistc.h"

#define EXTERN
EXTERN BOOL_VAR (tosp_old_to_method, FALSE, "Space stats use prechopping?");
EXTERN BOOL_VAR (tosp_only_use_prop_rows, TRUE,
"Block stats to use fixed pitch rows?");
EXTERN BOOL_VAR (tosp_use_pre_chopping, FALSE,
"Space stats use prechopping?");
EXTERN BOOL_VAR (tosp_old_to_bug_fix, FALSE, "Fix suspected bug in old code");
EXTERN BOOL_VAR (tosp_block_use_cert_spaces, TRUE,
"Only stat OBVIOUS spaces");
EXTERN BOOL_VAR (tosp_row_use_cert_spaces, TRUE, "Only stat OBVIOUS spaces");
EXTERN BOOL_VAR (tosp_narrow_blobs_not_cert, TRUE,
"Only stat OBVIOUS spaces");
EXTERN BOOL_VAR (tosp_row_use_cert_spaces1, TRUE, "Only stat OBVIOUS spaces");
EXTERN BOOL_VAR (tosp_recovery_isolated_row_stats, TRUE,
"Use row alone when inadequate cert spaces");
EXTERN BOOL_VAR (tosp_only_small_gaps_for_kern, FALSE, "Better guess");
EXTERN BOOL_VAR (tosp_all_flips_fuzzy, FALSE, "Pass ANY flip to context?");
EXTERN BOOL_VAR (tosp_fuzzy_limit_all, TRUE,
"Dont restrict kn->sp fuzzy limit to tables");
EXTERN BOOL_VAR (tosp_stats_use_xht_gaps, TRUE,
"Use within xht gap for wd breaks");
EXTERN BOOL_VAR (tosp_use_xht_gaps, TRUE, "Use within xht gap for wd breaks");
EXTERN BOOL_VAR (tosp_only_use_xht_gaps, FALSE,
"Only use within xht gap for wd breaks");
EXTERN BOOL_VAR (tosp_rule_9_test_punct, FALSE,
"Dont chng kn to space next to punct");
EXTERN BOOL_VAR (tosp_flip_fuzz_kn_to_sp, TRUE, "Default flip");
EXTERN BOOL_VAR (tosp_flip_fuzz_sp_to_kn, TRUE, "Default flip");
EXTERN BOOL_VAR (tosp_improve_thresh, FALSE, "Enable improvement heuristic");
EXTERN INT_VAR (tosp_debug_level, 0, "Debug data");
EXTERN INT_VAR (tosp_enough_space_samples_for_median, 3,
"or should we use mean");
EXTERN INT_VAR (tosp_redo_kern_limit, 10,
"No.samples reqd to reestimate for row");
EXTERN INT_VAR (tosp_few_samples, 40,
"No.gaps reqd with 1 large gap to treat as a table");
EXTERN INT_VAR (tosp_short_row, 20,
"No.gaps reqd with few cert spaces to use certs");
EXTERN INT_VAR (tosp_sanity_method, 1, "How to avoid being silly");
EXTERN double_VAR (tosp_threshold_bias1, 0,
"how far between kern and space?");
EXTERN double_VAR (tosp_threshold_bias2, 0,
"how far between kern and space?");
EXTERN double_VAR (tosp_narrow_fraction, 0.3, "Fract of xheight for narrow");
EXTERN double_VAR (tosp_narrow_aspect_ratio, 0.48,
"narrow if w/h less than this");
EXTERN double_VAR (tosp_wide_fraction, 0.52, "Fract of xheight for wide");
EXTERN double_VAR (tosp_wide_aspect_ratio, 0.0, "wide if w/h less than this");
EXTERN double_VAR (tosp_fuzzy_space_factor, 0.6,
"Fract of xheight for fuzz sp");
EXTERN double_VAR (tosp_fuzzy_space_factor1, 0.5,
"Fract of xheight for fuzz sp");
EXTERN double_VAR (tosp_fuzzy_space_factor2, 0.72,
"Fract of xheight for fuzz sp");
EXTERN double_VAR (tosp_gap_factor, 0.83, "gap ratio to flip sp->kern");
EXTERN double_VAR (tosp_kern_gap_factor1, 2.0, "gap ratio to flip kern->sp");
EXTERN double_VAR (tosp_kern_gap_factor2, 1.3, "gap ratio to flip kern->sp");
EXTERN double_VAR (tosp_kern_gap_factor3, 2.5, "gap ratio to flip kern->sp");
EXTERN double_VAR (tosp_ignore_big_gaps, -1, "xht multiplier");
EXTERN double_VAR (tosp_ignore_very_big_gaps, 3.5, "xht multiplier");
EXTERN double_VAR (tosp_rep_space, 1.6, "rep gap multiplier for space");
EXTERN double_VAR (tosp_enough_small_gaps, 0.65,
"Fract of kerns reqd for isolated row stats");
EXTERN double_VAR (tosp_table_kn_sp_ratio, 2.25,
"Min difference of kn & sp in table");
EXTERN double_VAR (tosp_table_xht_sp_ratio, 0.33,
"Expect spaces bigger than this");
EXTERN double_VAR (tosp_table_fuzzy_kn_sp_ratio, 3.0,
"Fuzzy if less than this");
EXTERN double_VAR (tosp_fuzzy_kn_fraction, 0.5, "New fuzzy kn alg");
EXTERN double_VAR (tosp_fuzzy_sp_fraction, 0.5, "New fuzzy sp alg");
EXTERN double_VAR (tosp_min_sane_kn_sp, 1.5,
"Dont trust spaces less than this time kn");
EXTERN double_VAR (tosp_init_guess_kn_mult, 2.2,
"Thresh guess - mult kn by this");
EXTERN double_VAR (tosp_init_guess_xht_mult, 0.28,
"Thresh guess - mult xht by this");
EXTERN double_VAR (tosp_max_sane_kn_thresh, 5.0,
"Multiplier on kn to limit thresh");
EXTERN double_VAR (tosp_flip_caution, 0.0,
"Dont autoflip kn to sp when large separation");

EXTERN double_VAR (tosp_large_kerning, 0.19,
"Limit use of xht gap with large kns");
EXTERN double_VAR (tosp_dont_fool_with_small_kerns, -1,
"Limit use of xht gap with odd small kns");
EXTERN double_VAR (tosp_near_lh_edge, 0,
"Dont reduce box if the top left is non blank");
EXTERN double_VAR (tosp_silly_kn_sp_gap, 0.2,
"Dont let sp minus kn get too small");
EXTERN double_VAR (tosp_pass_wide_fuzz_sp_to_context, 0.75,
"How wide fuzzies need context");

#define MAXSPACING      128      /*max expected spacing in pix */
/**********************************************************************
 * to_spacing
 *
 * Compute fuzzy word spacing thresholds for each row.
 * I.e. set :   max_nonspace
 *							space_threshold
 *							min_space
 *							kern_size
 *							space_size     for each row.
 * ONLY FOR PROPORTIONAL BLOCKS - FIXED PITCH IS ASSUMED ALREADY DONE
 **********************************************************************/

void to_spacing(                       //set spacing
                ICOORD page_tr,        //topright of page
                TO_BLOCK_LIST *blocks  //blocks on page
               ) {
  TO_BLOCK_IT block_it;          //iterator
  TO_BLOCK *block;               //current block;
  TO_ROW_IT row_it;              //row iterator
  TO_ROW *row;                   //current row
  int block_index;               //block number
  int row_index;                 //row number
  INT16 block_space_gap_width;   //Estimated width of    real spaces for whole block
                                 //Estimate width ofnon space gaps for whole block
  INT16 block_non_space_gap_width;
                                 //Old fixed/prop result
  BOOL8 old_text_ord_proportional;
  GAPMAP *gapmap = NULL;         //map of big vert gaps in blk

  block_it.set_to_list (blocks);
  block_index = 1;
  for (block_it.mark_cycle_pt (); !block_it.cycled_list ();
  block_it.forward ()) {
    block = block_it.data ();
    gapmap = new GAPMAP (block);
    block_spacing_stats(block,
                        gapmap,
                        old_text_ord_proportional,
                        block_space_gap_width,
                        block_non_space_gap_width);
    row_it.set_to_list (block->get_rows ());
    row_index = 1;
    for (row_it.mark_cycle_pt (); !row_it.cycled_list (); row_it.forward ()) {
      row = row_it.data ();
      if ((row->pitch_decision == PITCH_DEF_PROP) ||
      (row->pitch_decision == PITCH_CORR_PROP)) {
        if ((tosp_debug_level > 0) && !old_text_ord_proportional)
          tprintf ("Block %d Row %d: Now Proportional\n",
            block_index, row_index);
        row_spacing_stats(row,
                          gapmap,
                          block_index,
                          row_index,
                          block_space_gap_width,
                          block_non_space_gap_width);
      }
      else {
        if ((tosp_debug_level > 0) && old_text_ord_proportional)
          tprintf
            ("Block %d Row %d: Now Fixed Pitch Decision:%d fp flag:%f\n",
            block_index, row_index, row->pitch_decision,
            row->fixed_pitch);
      }
#ifndef GRAPHICS_DISABLED
      if (textord_show_initial_words)
        plot_word_decisions (to_win, (INT16) row->fixed_pitch, row);
#endif
      row_index++;
    }
    delete gapmap;
    block_index++;
  }
}


/*************************************************************************
 * block_spacing_stats()
 *************************************************************************/

void block_spacing_stats(                                  //DEBUG USE ONLY
                         TO_BLOCK *block,
                         GAPMAP *gapmap,
                         BOOL8 &old_text_ord_proportional,
                         INT16 &block_space_gap_width,     //resulting estimate
                         INT16 &block_non_space_gap_width  //resulting estimate
                        ) {
  TO_ROW_IT row_it;              //row iterator
  TO_ROW *row;                   //current row
  BLOBNBOX_IT blob_it;           //iterator

  STATS centre_to_centre_stats (0, MAXSPACING);
  //DEBUG USE ONLY
  STATS all_gap_stats (0, MAXSPACING);
  STATS space_gap_stats (0, MAXSPACING);
  INT16 minwidth = MAX_INT16;    //narrowest blob
  BOX blob_box;
  BOX prev_blob_box;
  INT16 centre_to_centre;
  INT16 gap_width;
  float real_space_threshold;
  float iqr_centre_to_centre;    //DEBUG USE ONLY
  float iqr_all_gap_stats;       //DEBUG USE ONLY
  INT32 end_of_row;
  INT32 row_length;

  row_it.set_to_list (block->get_rows ());
  for (row_it.mark_cycle_pt (); !row_it.cycled_list (); row_it.forward ()) {
    row = row_it.data ();
    if (!row->blob_list ()->empty () &&
      (!tosp_only_use_prop_rows ||
      (row->pitch_decision == PITCH_DEF_PROP) ||
    (row->pitch_decision == PITCH_CORR_PROP))) {
      blob_it.set_to_list (row->blob_list ());
      blob_it.mark_cycle_pt ();
      end_of_row = blob_it.data_relative (-1)->bounding_box ().right ();
      if (tosp_use_pre_chopping)
        blob_box = box_next_pre_chopped (&blob_it);
      else if (tosp_stats_use_xht_gaps)
        blob_box = reduced_box_next (row, &blob_it);
      else
        blob_box = box_next (&blob_it);
      row_length = end_of_row - blob_box.left ();
      if (blob_box.width () < minwidth)
        minwidth = blob_box.width ();
      prev_blob_box = blob_box;
      while (!blob_it.cycled_list ()) {
        if (tosp_use_pre_chopping)
          blob_box = box_next_pre_chopped (&blob_it);
        else if (tosp_stats_use_xht_gaps)
          blob_box = reduced_box_next (row, &blob_it);
        else
          blob_box = box_next (&blob_it);
        if (blob_box.width () < minwidth)
          minwidth = blob_box.width ();
        gap_width = blob_box.left () - prev_blob_box.right ();
        if (!ignore_big_gap (row, row_length, gapmap,
        prev_blob_box.right (), blob_box.left ())) {
          all_gap_stats.add (gap_width, 1);

          centre_to_centre = (blob_box.left () + blob_box.right () -
            (prev_blob_box.left () +
            prev_blob_box.right ())) / 2;
          //DEBUG
          centre_to_centre_stats.add (centre_to_centre, 1);
          // DEBUG
        }
        prev_blob_box = blob_box;
      }
    }
  }

                                 //Inadequate samples
  if (all_gap_stats.get_total () <= 1) {
    block_non_space_gap_width = minwidth;
    block_space_gap_width = -1;  //No est. space width
                                 //DEBUG
    old_text_ord_proportional = TRUE;
  }
  else {
    /* For debug only ..... */
    iqr_centre_to_centre = centre_to_centre_stats.ile (0.75) -
      centre_to_centre_stats.ile (0.25);
    iqr_all_gap_stats = all_gap_stats.ile (0.75) - all_gap_stats.ile (0.25);
    old_text_ord_proportional =
      iqr_centre_to_centre * 2 > iqr_all_gap_stats;
    /* .......For debug only */

    /*
    The median of the gaps is used as an estimate of the NON-SPACE gap width.
    This RELIES on the assumption that there are more gaps WITHIN words than
    BETWEEN words in a block

    Now try to estimate the width of a real space for all real spaces in the
    block. Do this by using a crude threshold to ignore "narrow" gaps, then
    find the median of the "wide" gaps and use this.
    */
    block_non_space_gap_width = (INT16) floor (all_gap_stats.median ());
    // median gap

    row_it.set_to_list (block->get_rows ());
    for (row_it.mark_cycle_pt (); !row_it.cycled_list (); row_it.forward ()) {
      row = row_it.data ();
      if (!row->blob_list ()->empty () &&
        (!tosp_only_use_prop_rows ||
        (row->pitch_decision == PITCH_DEF_PROP) ||
      (row->pitch_decision == PITCH_CORR_PROP))) {
        real_space_threshold =
          MAX (tosp_init_guess_kn_mult * block_non_space_gap_width,
          tosp_init_guess_xht_mult * row->xheight);
        blob_it.set_to_list (row->blob_list ());
        blob_it.mark_cycle_pt ();
        end_of_row =
          blob_it.data_relative (-1)->bounding_box ().right ();
        if (tosp_use_pre_chopping)
          blob_box = box_next_pre_chopped (&blob_it);
        else if (tosp_stats_use_xht_gaps)
          blob_box = reduced_box_next (row, &blob_it);
        else
          blob_box = box_next (&blob_it);
        row_length = blob_box.left () - end_of_row;
        prev_blob_box = blob_box;
        while (!blob_it.cycled_list ()) {
          if (tosp_use_pre_chopping)
            blob_box = box_next_pre_chopped (&blob_it);
          else if (tosp_stats_use_xht_gaps)
            blob_box = reduced_box_next (row, &blob_it);
          else
            blob_box = box_next (&blob_it);
          gap_width = blob_box.left () - prev_blob_box.right ();
          if ((gap_width > real_space_threshold) &&
            !ignore_big_gap (row, row_length, gapmap,
            prev_blob_box.right (),
          blob_box.left ())) {
            /*
            If tosp_use_cert_spaces is enabled, the estimate of the space gap is
            restricted to obvious spaces - those wider than half the xht or those
            with wide blobs on both sides - i.e not things that are suspect 1's or
            punctiation that is sometimes widely spaced.
            */
            if (!tosp_block_use_cert_spaces ||
              (gap_width >
              tosp_fuzzy_space_factor2 * row->xheight)
              ||
              ((gap_width >
              tosp_fuzzy_space_factor1 * row->xheight)
              && (!tosp_narrow_blobs_not_cert
              || (!narrow_blob (row, prev_blob_box)
              && !narrow_blob (row, blob_box))))
              || (wide_blob (row, prev_blob_box)
              && wide_blob (row, blob_box)))
              space_gap_stats.add (gap_width, 1);
          }
          prev_blob_box = blob_box;
        }
      }
    }
                                 //Inadequate samples
    if (space_gap_stats.get_total () <= 2)
      block_space_gap_width = -1;//No est. space width
    else
      block_space_gap_width =
        MAX ((INT16) floor (space_gap_stats.median ()),
        3 * block_non_space_gap_width);
  }
}


/*************************************************************************
 * row_spacing_stats()
 * Set values for min_space, max_non_space based on row stats only
 * If failure - return 0 values.
 *************************************************************************/

void row_spacing_stats(                                 //estimate for block
                       TO_ROW *row,
                       GAPMAP *gapmap,
                       INT16 block_idx,
                       INT16 row_idx,
                       INT16 block_space_gap_width,
                       INT16 block_non_space_gap_width  //estimate for block
                      ) {