makerow.cpp

来自「一ＯＣＲ的相关资料。．希望对研究ＯＣＲ的朋友有所帮助．」· C++ 代码 · 共 1,691 行 · 第 1/5 页

      * (textord_merge_x + textord_merge_asc);
    if (y_min > y_bottom) {      //expansion allowed
                                 //expandable
      swallowed_row = TRUE;
      while (swallowed_row && !row_it.at_last ()) {
        swallowed_row = FALSE;
                                 //get next one
        test_row = row_it.data_relative (1);
                                 //overlaps space
        if (test_row->max_y () > y_bottom) {
          if (test_row->min_y () > y_bottom) {
            row_it.forward ();
#ifndef GRAPHICS_DISABLED
            if (textord_show_expanded_rows && testing_on)
              plot_parallel_row(test_row,
                                gradient,
                                block_edge,
                                WHITE,
                                rotation);
#endif
            blob_it.set_to_list (row->blob_list ());
            blob_it.add_list_after (test_row->blob_list ());
                                 //swallow complete row
            delete row_it.extract ();
            row_it.backward ();
            swallowed_row = TRUE;
          }
          else if (test_row->max_y () < y_min)
                                 //shorter limit
            y_bottom = test_row->max_y ();
          else
            y_bottom = y_min;    //can't expand it
        }
      }
      y_min = y_bottom;          //expand it
    }
    if (y_max < y_top) {         //expansion allowed
      swallowed_row = TRUE;
      while (swallowed_row && !row_it.at_first ()) {
        swallowed_row = FALSE;
                                 //get one above
        test_row = row_it.data_relative (-1);
        if (test_row->min_y () < y_top) {
          if (test_row->max_y () < y_top) {
            row_it.backward ();
            blob_it.set_to_list (row->blob_list ());
#ifndef GRAPHICS_DISABLED
            if (textord_show_expanded_rows && testing_on)
              plot_parallel_row(test_row,
                                gradient,
                                block_edge,
                                WHITE,
                                rotation);
#endif
            blob_it.add_list_after (test_row->blob_list ());
                                 //swallow complete row
            delete row_it.extract ();
            row_it.forward ();
            swallowed_row = TRUE;
          }
          else if (test_row->min_y () < y_max)
                                 //shorter limit
            y_top = test_row->min_y ();
          else
            y_top = y_max;       //can't expand it

        }
      }
      y_max = y_top;
    }
                                 //new limits
    row->set_limits (y_min, y_max);
    row_it.backward ();
  }
  while (!row_it.at_last ());
}


/**********************************************************************
 * adjust_row_limits
 *
 * Change the limits of rows to suit the default fractions.
 **********************************************************************/

void adjust_row_limits(                 //tidy limits
                       TO_BLOCK *block  //block to do
                      ) {
  TO_ROW *row;                   //current row
  float size;                    //size of row
  float ymax;                    //top of row
  float ymin;                    //bottom of row
  TO_ROW_IT row_it = block->get_rows ();

  for (row_it.mark_cycle_pt (); !row_it.cycled_list (); row_it.forward ()) {
    row = row_it.data ();
    size = row->max_y () - row->min_y ();
    size /= textord_merge_x + textord_merge_asc + textord_merge_desc;
    ymax = size * (textord_merge_x + textord_merge_asc);
    ymin = -size * textord_merge_desc;
    row->set_limits (row->intercept () + ymin, row->intercept () + ymax);
    row->merged = FALSE;
  }
}


/**********************************************************************
 * compute_row_stats
 *
 * Compute the linespacing and offset.
 **********************************************************************/

void compute_row_stats(                  //find lines
                       TO_BLOCK *block,  //block to do
                       BOOL8 testing_on  //correct orientation
                      ) {
  INT32 row_index;               //of median
  TO_ROW *row;                   //current row
  TO_ROW *prev_row;              //previous row
  float iqr;                     //inter quartile range
  TO_ROW_IT row_it = block->get_rows ();
                                 //number of rows
  INT16 rowcount = row_it.length ();
  TO_ROW **rows;                 //for choose nth

  rows = (TO_ROW **) alloc_mem (rowcount * sizeof (TO_ROW *));
  if (rows == NULL)
    MEMORY_OUT.error ("compute_row_stats", ABORT, NULL);
  rowcount = 0;
  prev_row = NULL;
  row_it.move_to_last ();        //start at bottom
  do {
    row = row_it.data ();
    if (prev_row != NULL) {
      rows[rowcount++] = prev_row;
      prev_row->spacing = row->intercept () - prev_row->intercept ();
      if (testing_on)
        tprintf ("Row at %g yields spacing of %g\n",
          row->intercept (), prev_row->spacing);
    }
    prev_row = row;
    row_it.backward ();
  }
  while (!row_it.at_last ());
  block->key_row = prev_row;
  block->baseline_offset =
    fmod (prev_row->parallel_c (), block->line_spacing);
  if (testing_on)
    tprintf ("Blob based spacing=(%g,%g), offset=%g",
      block->line_size, block->line_spacing, block->baseline_offset);
  if (rowcount > 0) {
    row_index = choose_nth_item (rowcount * 3 / 4, rows, rowcount,
      sizeof (TO_ROW *), row_spacing_order);
    iqr = rows[row_index]->spacing;
    row_index = choose_nth_item (rowcount / 4, rows, rowcount,
      sizeof (TO_ROW *), row_spacing_order);
    iqr -= rows[row_index]->spacing;
    row_index = choose_nth_item (rowcount / 2, rows, rowcount,
      sizeof (TO_ROW *), row_spacing_order);
    block->key_row = rows[row_index];
    if (testing_on)
      tprintf (" row based=%g(%g)", rows[row_index]->spacing, iqr);
    if (rowcount > 2
    && iqr < rows[row_index]->spacing * textord_linespace_iqrlimit) {
      if (!textord_new_initial_xheight) {
        if (rows[row_index]->spacing < block->line_spacing
          && rows[row_index]->spacing > block->line_size)
          //within range
          block->line_size = rows[row_index]->spacing;
        //spacing=size
        else if (rows[row_index]->spacing > block->line_spacing)
          block->line_size = block->line_spacing;
        //too big so use max
      }
      else {
        if (rows[row_index]->spacing < block->line_spacing)
          block->line_size = rows[row_index]->spacing;
        else
          block->line_size = block->line_spacing;
        //too big so use max
      }
      if (block->line_size < textord_min_xheight)
        block->line_size = (float) textord_min_xheight;
      block->line_spacing = rows[row_index]->spacing;
      block->max_blob_size =
        block->line_spacing * textord_excess_blobsize;
    }
    block->baseline_offset = fmod (rows[row_index]->intercept (),
      block->line_spacing);
  }
  if (testing_on)
    tprintf ("\nEstimate line size=%g, spacing=%g, offset=%g\n",
      block->line_size, block->line_spacing, block->baseline_offset);
  free_mem(rows);
}


/**********************************************************************
 * compute_block_xheight
 *
 * Compute the xheight of the individual rows, then correlate them
 * and interpret ascenderless lines, correcting xheights.
 **********************************************************************/

void compute_block_xheight(                  //find lines
                           TO_BLOCK *block,  //block to do
                           float gradient    //global skew
                          ) {
  TO_ROW *row;                   //current row
  int xh_count, desc_count;      //no of samples
  float block_median;            //median blob size
  int asc_count, cap_count;
  INT32 min_size, max_size;      //limits on xheight
  INT32 evidence;                //no of samples on row
  float xh_sum, desc_sum;        //for averages
  float asc_sum, cap_sum;
  TO_ROW_IT row_it = block->get_rows ();
  STATS row_heights;             //block evidence

  if (row_it.empty ())
    return;                      //no rows
  block_median = median_block_xheight (block, gradient);
  block_median *= 2;
  if (block_median < block->line_size)
    block_median = block->line_size;
  //      tprintf("Block median=%g, linesize=%g\n",
  //              block_median,block->line_size);
  max_size = (INT32) ceil (block_median);
  min_size = (INT32) floor (block_median * textord_minxh);
  row_heights.set_range (min_size, max_size + 1);
  xh_count = desc_count = asc_count = cap_count = 0;
  xh_sum = desc_sum = asc_sum = cap_sum = 0.0f;
  for (row_it.mark_cycle_pt (); !row_it.cycled_list (); row_it.forward ()) {
    row = row_it.data ();
    evidence = compute_row_xheight (row, min_size, max_size, gradient);
    if (row->xheight > 0 && row->ascrise > 0) {
      row_heights.add ((INT32) row->xheight, evidence);
      xh_count += evidence;
      asc_sum += row->ascrise;
      asc_count++;
    }
    else if (row->xheight > 0) {
      cap_sum += row->xheight;   //assume just caps
      cap_count++;
    }
    if (row->descdrop != 0) {
      desc_sum += row->descdrop;
      desc_count++;
    }
  }
  if (xh_count > 0) {
                                 //median
    xh_sum = row_heights.ile (0.5);
    asc_sum /= asc_count;
  }
  else if (cap_count > 0) {
    cap_sum /= cap_count;        //must assume caps
    xh_sum =
      cap_sum * textord_merge_x / (textord_merge_x + textord_merge_asc);
    asc_sum =
      cap_sum * textord_merge_asc / (textord_merge_x + textord_merge_asc);
  }
  else {
                                 //default sizes
    xh_sum = block_median * textord_merge_x;
    asc_sum = block_median * textord_merge_asc;
  }
  if (desc_count > 0) {
    desc_sum /= desc_count;
  }
  else {
    desc_sum = xh_sum * textord_merge_desc / textord_merge_x;
  }
  //      tprintf("Block average x height=%g, count=%d, asc=%g/%d, desc=%g/%d,cap=%g/%d\n",
  //              xh_sum,xh_count,asc_sum,asc_count,desc_sum,desc_count,
  //              cap_sum,cap_count);
  if (xh_sum < textord_min_xheight)
    xh_sum = (float) textord_min_xheight;
  block->xheight = xh_sum;
  for (row_it.mark_cycle_pt (); !row_it.cycled_list (); row_it.forward ()) {
    correct_row_xheight (row_it.data (), xh_sum, asc_sum, desc_sum);
  }
}


/**********************************************************************
 * median_block_xheight
 *
 * Compute the linespacing and offset.
 **********************************************************************/

float median_block_xheight(                  //find lines
                           TO_BLOCK *block,  //block to do
                           float gradient    //global skew
                          ) {
  TO_ROW *row;                   //current row
  float result;                  //output size
  float xcentre;                 //centre of blob
  TO_ROW_IT row_it = block->get_rows ();
  BLOBNBOX_IT blob_it;
  BLOBNBOX *blob;                //current blob
  float *heights;                //for choose nth
  INT32 blob_count;              //blobs in block
  INT32 blob_index;              //current blob

  blob_count = 0;
  for (row_it.mark_cycle_pt (); !row_it.cycled_list (); row_it.forward ())
    blob_count += row_it.data ()->blob_list ()->length ();
  heights = (float *) alloc_mem (blob_count * sizeof (float));
  if (heights == NULL)
    MEMORY_OUT.error ("compute_row_stats", ABORT, NULL);

  blob_index = 0;
  for (row_it.mark_cycle_pt (); !row_it.cycled_list (); row_it.forward ()) {
    row = row_it.data ();
    blob_it.set_to_list (row->blob_list ());
    for (blob_it.mark_cycle_pt (); !blob_it.cycled_list ();
    blob_it.forward ()) {
      blob = blob_it.data ();
      if (!blob->joined_to_prev ()) {
        xcentre =
          (blob->bounding_box ().left () +
          blob->bounding_box ().right ()) / 2.0f;
        heights[blob_index] =
          blob->bounding_box ().top () - gradient * xcentre -
          row->parallel_c ();
        if (heights[blob_index] > 0)
          blob_index++;
      }
    }
  }
  ASSERT_HOST (blob_index > 0);  //dont expect 0
  blob_count = blob_index;
  blob_index = choose_nth_item (blob_count / 2, heights, blob_count);
  result = heights[blob_index];
  free_mem(heights);
  return result;
}