docqual.cpp

一ＯＣＲ的相关资料。．希望对研究ＯＣＲ的朋友有所帮助．

  delete_word(tessword);  //get rid of it

  initial_it.set_to_list (init_word->blob_list ());
  init_blobs_left = initial_it.length ();
  outword_it.set_to_list (word->outword->blob_list ());

  for (outword_it.mark_cycle_pt ();
  !outword_it.cycled_list (); outword_it.forward ()) {
    out_box = outword_it.data ()->bounding_box ();

    /* Skip any initial blobs LEFT of current outword blob */
    while (!initial_it.at_last () &&
    (initial_it.data ()->bounding_box ().left () < out_box.left ())) {
      initial_it.forward ();
      init_blobs_left--;
    }

    /* See if current outword blob matches any initial blob with the same left
      coord. (Normally only one but possibly more - in unknown order) */

    i = 0;
    matched = FALSE;
    do {
      test_blob = initial_it.data_relative (i++);
      matched = crude_match_blobs (test_blob, outword_it.data ());
      if (matched &&
        (word->reject_map[j].accept_if_good_quality ()) &&
        (docqual_excuse_outline_errs ||
        (count_outline_errs (word->best_choice->string ()[j],
        outword_it.data ()->out_list ()->
        length ()) == 0)))
        word->reject_map[j].setrej_quality_accept ();
    }
    while (!matched &&
      (init_blobs_left - i > 0) &&
      (i < 129) &&
      !initial_it.at_last () &&
      test_blob->bounding_box ().left () == out_box.left ());
    j++;
  }
  delete init_word;
}


void print_boxes(WERD *word) { 
  PBLOB_IT it;
  BOX box;

  it.set_to_list (word->blob_list ());
  for (it.mark_cycle_pt (); !it.cycled_list (); it.forward ()) {
    box = it.data ()->bounding_box ();
    box.print ();
  }
}


INT16 count_outline_errs(char c, INT16 outline_count) { 
  int expected_outline_count;

  if (STRING (outlines_odd).contains (c))
    return 0;                    //Dont use this char
  else if (STRING (outlines_2).contains (c))
    expected_outline_count = 2;
  else
    expected_outline_count = 1;
  return abs (outline_count - expected_outline_count);
}


void quality_based_rejection(PAGE_RES_IT &page_res_it,
                             BOOL8 good_quality_doc) {
  if ((tessedit_good_quality_unrej && good_quality_doc))
    unrej_good_quality_words(page_res_it); 
  doc_and_block_rejection(page_res_it, good_quality_doc); 

  page_res_it.restart_page ();
  while (page_res_it.word () != NULL) {
    insert_rej_cblobs (page_res_it.word ());
    page_res_it.forward ();
  }

  if (unlv_tilde_crunching) {
    tilde_crunch(page_res_it); 
    tilde_delete(page_res_it); 
  }
}


/*************************************************************************
 * unrej_good_quality_words()
 * Accept potential rejects in words which pass the following checks:
 *    - Contains a potential reject
 *    - Word looks like a sensible alpha word.
 *    - Word segmentation is the same as the original image
 *		- All characters have the expected number of outlines
 * NOTE - the rejection counts are recalculated after unrejection
 *      - CANT do it in a single pass without a bit of fiddling
 *		- keep it simple but inefficient
 *************************************************************************/
void unrej_good_quality_words(  //unreject potential
                              PAGE_RES_IT &page_res_it) {
  WERD_RES *word;
  ROW_RES *current_row;
  BLOCK_RES *current_block;
  int i;

  page_res_it.restart_page ();
  while (page_res_it.word () != NULL) {
    check_debug_pt (page_res_it.word (), 100);
    if (bland_unrej) {
      word = page_res_it.word ();
      for (i = 0; i < word->reject_map.length (); i++) {
        if (word->reject_map[i].accept_if_good_quality ())
          word->reject_map[i].setrej_quality_accept ();
      }
      page_res_it.forward ();
    }
    else if ((page_res_it.row ()->char_count > 0) &&
      ((page_res_it.row ()->rej_count /
      (float) page_res_it.row ()->char_count) <=
    quality_rowrej_pc)) {
      word = page_res_it.word ();
      if (word->reject_map.quality_recoverable_rejects () &&
        (tessedit_unrej_any_wd ||
        acceptable_word_string (word->best_choice->string ().string ())
      != AC_UNACCEPTABLE)) {
        unrej_good_chs (word, page_res_it.row ()->row);
      }
      page_res_it.forward ();
    }
    else {
      /* Skip to end of dodgy row */
      current_row = page_res_it.row ();
      while ((page_res_it.word () != NULL) &&
        (page_res_it.row () == current_row))
        page_res_it.forward ();
    }
    check_debug_pt (page_res_it.word (), 110);
  }
  page_res_it.restart_page ();
  page_res_it.page_res->char_count = 0;
  page_res_it.page_res->rej_count = 0;
  current_block = NULL;
  current_row = NULL;
  while (page_res_it.word () != NULL) {
    if (current_block != page_res_it.block ()) {
      current_block = page_res_it.block ();
      current_block->char_count = 0;
      current_block->rej_count = 0;
    }
    if (current_row != page_res_it.row ()) {
      current_row = page_res_it.row ();
      current_row->char_count = 0;
      current_row->rej_count = 0;
      current_row->whole_word_rej_count = 0;
    }
    page_res_it.rej_stat_word ();
    page_res_it.forward ();
  }
}


/*************************************************************************
 * doc_and_block_rejection()
 *
 * If the page has too many rejects - reject all of it.
 * If any block has too many rejects - reject all words in the block
 *************************************************************************/

void doc_and_block_rejection(  //reject big chunks
                             PAGE_RES_IT &page_res_it,
                             BOOL8 good_quality_doc) {
  INT16 block_no = 0;
  INT16 row_no = 0;
  BLOCK_RES *current_block;
  ROW_RES *current_row;

  BOOL8 rej_word;
  BOOL8 prev_word_rejected;
  INT16 char_quality;
  INT16 accepted_char_quality;

  if ((page_res_it.page_res->rej_count * 100.0 /
  page_res_it.page_res->char_count) > tessedit_reject_doc_percent) {
    reject_whole_page(page_res_it); 
    #ifndef SECURE_NAMES
    if (tessedit_debug_doc_rejection) {
      tprintf ("REJECT ALL #chars: %d #Rejects: %d; \n",
        page_res_it.page_res->char_count,
        page_res_it.page_res->rej_count);
    }
    #endif
  }
  else {
    #ifndef SECURE_NAMES
    if (tessedit_debug_doc_rejection)
      tprintf ("NO PAGE REJECTION #chars: %d  # Rejects: %d; \n",
        page_res_it.page_res->char_count,
        page_res_it.page_res->rej_count);
    #endif

    /* Walk blocks testing for block rejection */

    page_res_it.restart_page ();
    while (page_res_it.word () != NULL) {
      current_block = page_res_it.block ();
      if (current_block->block->text_region () != NULL)
        block_no = current_block->block->text_region ()->id_no ();
      else
        block_no = -1;
      if ((page_res_it.block ()->char_count > 0) &&
        ((page_res_it.block ()->rej_count * 100.0 /
        page_res_it.block ()->char_count) >
      tessedit_reject_block_percent)) {
        #ifndef SECURE_NAMES
        if (tessedit_debug_block_rejection)
          tprintf ("REJECTING BLOCK %d  #chars: %d;  #Rejects: %d\n",
            block_no,
            page_res_it.block ()->char_count,
            page_res_it.block ()->rej_count);
        #endif
        prev_word_rejected = FALSE;
        while ((page_res_it.word () != NULL) &&
        (page_res_it.block () == current_block)) {
          if (tessedit_preserve_blk_rej_perfect_wds) {
            rej_word =
              (page_res_it.word ()->reject_map.reject_count () > 0)
              || (page_res_it.word ()->reject_map.length () <
              tessedit_preserve_min_wd_len);
            if (rej_word && tessedit_dont_blkrej_good_wds
              && !(page_res_it.word ()->reject_map.length () <
              tessedit_preserve_min_wd_len)
              &&
              (acceptable_word_string
              (page_res_it.word ()->best_choice->string ().
            string ()) != AC_UNACCEPTABLE)) {
              word_char_quality (page_res_it.word (),
                page_res_it.row ()->row,
                &char_quality,
                &accepted_char_quality);
              rej_word = char_quality !=
                page_res_it.word ()->reject_map.length ();
            }
          }
          else
            rej_word = TRUE;
          if (rej_word) {
            /*
              Reject spacing if both current and prev words are rejected.
              NOTE - this is NOT restricted to FUZZY spaces. - When tried this generated
              more space errors.
            */
            if (tessedit_use_reject_spaces &&
              prev_word_rejected &&
              (page_res_it.prev_row () == page_res_it.row ()) &&
              (page_res_it.word ()->word->space () == 1))
              page_res_it.word ()->reject_spaces = TRUE;
            page_res_it.word ()->reject_map.rej_word_block_rej ();
          }
          prev_word_rejected = rej_word;
          page_res_it.forward ();
        }
      }
      else {
        #ifndef SECURE_NAMES
        if (tessedit_debug_block_rejection)
          tprintf
            ("NOT REJECTING BLOCK %d #chars: %d  # Rejects: %d; \n",
            block_no, page_res_it.block ()->char_count,
            page_res_it.block ()->rej_count);
        #endif

        /* Walk rows in block testing for row rejection */
        row_no = 0;
        while ((page_res_it.word () != NULL) &&
        (page_res_it.block () == current_block)) {
          current_row = page_res_it.row ();
          row_no++;
          /* Reject whole row if:
            fraction of chars on row which are rejected exceed a limit AND
            fraction rejects which occur in WHOLE WERD rejects is LESS THAN a limit
          */
          if ((page_res_it.row ()->char_count > 0) &&
            ((page_res_it.row ()->rej_count * 100.0 /
            page_res_it.row ()->char_count) >
            tessedit_reject_row_percent) &&
            ((page_res_it.row ()->whole_word_rej_count * 100.0 /
            page_res_it.row ()->rej_count) <
          tessedit_whole_wd_rej_row_percent)) {
            #ifndef SECURE_NAMES
            if (tessedit_debug_block_rejection)
              tprintf
                ("REJECTING ROW %d  #chars: %d;  #Rejects: %d\n",
                row_no, page_res_it.row ()->char_count,
                page_res_it.row ()->rej_count);
            #endif
            prev_word_rejected = FALSE;
            while ((page_res_it.word () != NULL) &&
            (page_res_it.row () == current_row)) {
              /* Preserve words on good docs unless they are mostly rejected*/
              if (!tessedit_row_rej_good_docs && good_quality_doc) {
                rej_word =
                  page_res_it.word ()->reject_map.
                  reject_count () /
                  (float) page_res_it.word ()->reject_map.
                  length () > tessedit_good_doc_still_rowrej_wd;
              }

              /* Preserve perfect words anyway */
              else if (tessedit_preserve_row_rej_perfect_wds) {
                rej_word =
                  (page_res_it.word ()->reject_map.
                  reject_count () > 0)
                  || (page_res_it.word ()->reject_map.
                  length () < tessedit_preserve_min_wd_len);
                if (rej_word && tessedit_dont_rowrej_good_wds
                  && !(page_res_it.word ()->reject_map.
                  length () <
                  tessedit_preserve_min_wd_len)
                  &&
                  (acceptable_word_string
                  (page_res_it.word ()->best_choice->
                string ().string ()) != AC_UNACCEPTABLE)) {
                  word_char_quality (page_res_it.word (),
                    page_res_it.row ()->row,
                    &char_quality,
                    &accepted_char_quality);
                  rej_word = char_quality !=
                    page_res_it.word ()->reject_map.length ();
                }
              }
              else
                rej_word = TRUE;
              if (rej_word) {
                /*
                  Reject spacing if both current and prev words are rejected.
                  NOTE - this is NOT restricted to FUZZY spaces. - When tried this generated
                  more space errors.
                */
                if (tessedit_use_reject_spaces &&
                  prev_word_rejected &&
                  (page_res_it.prev_row () ==
                  page_res_it.row ())
                  && (page_res_it.word ()->word->space () ==
                  1))
                  page_res_it.word ()->reject_spaces = TRUE;
                page_res_it.word ()->reject_map.
                  rej_word_row_rej(); 
              }
              prev_word_rejected = rej_word;
              page_res_it.forward ();
            }
          }
          else {
            #ifndef SECURE_NAMES
            if (tessedit_debug_block_rejection)
              tprintf
                ("NOT REJECTING ROW %d #chars: %d  # Rejects: %d; \n",
                row_no, page_res_it.row ()->char_count,
                page_res_it.row ()->rej_count);
            #endif
            while ((page_res_it.word () != NULL) &&
              (page_res_it.row () == current_row))
              page_res_it.forward ();