adaptions.cpp

一ＯＣＲ的相关资料。．希望对研究ＯＣＲ的朋友有所帮助．

        #ifndef SECURE_NAMES
        if (tessedit_cluster_debug)
          tprintf ("Sample added to an existing cluster\n");
        #endif
      }
      else {
        #ifndef SECURE_NAMES
        if (tessedit_cluster_debug)
          tprintf
            ("Sample dropped, good match to an existing cluster\n");
        #endif
      }
    }
    else if (best_score > tessedit_cluster_t2) {
      c_it.add_to_end (new CHAR_SAMPLES (sample));
      #ifndef SECURE_NAMES
      if (tessedit_cluster_debug)
        tprintf ("New cluster created for this sample\n");
      #endif
    }
    else {
      cw_it.add_to_end (sample);
      if (tessedit_cluster_debug)
        tprintf ("Sample added to the wait list\n");
    }
  }
}


void check_wait_list(CHAR_SAMPLE_LIST *chars_waiting,
                     CHAR_SAMPLE *sample,
                     CHAR_SAMPLES *best_cluster) {
  CHAR_SAMPLE *wait_sample;
  CHAR_SAMPLE *test_sample = sample;
  CHAR_SAMPLE_IT cw_it = chars_waiting;
  CHAR_SAMPLE_LIST add_list;     //Samples added to best cluster
  CHAR_SAMPLE_IT add_it = &add_list;
  float score;

  add_list.clear ();

  if (!cw_it.empty ()) {
    do {
      if (!add_list.empty ()) {
        add_it.forward ();
        test_sample = add_it.extract ();
        best_cluster->add_sample (test_sample);
      }

      for (cw_it.mark_cycle_pt ();
      !cw_it.cycled_list (); cw_it.forward ()) {
        wait_sample = cw_it.data ();
        if (tessedit_mm_use_prototypes)
          score = best_cluster->match_score (wait_sample);
        else
          score = sample->match_sample (wait_sample, FALSE);
        if (score < tessedit_cluster_t1) {
          if (score > tessedit_cluster_t3
          || tessedit_mm_use_prototypes) {
            add_it.add_after_stay_put (cw_it.extract ());
            #ifndef SECURE_NAMES
            if (tessedit_cluster_debug)
              tprintf
                ("Wait sample added to an existing cluster\n");
            #endif
          }
          else {
            #ifndef SECURE_NAMES
            if (tessedit_cluster_debug)
              tprintf
                ("Wait sample dropped, good match to an existing cluster\n");
            #endif
          }
        }
      }
    }
    while (!add_list.empty ());
  }
}


void complete_clustering(CHAR_SAMPLES_LIST *char_clusters,
                         CHAR_SAMPLE_LIST *chars_waiting) {
  CHAR_SAMPLES *best_cluster;
  CHAR_SAMPLES_IT c_it = char_clusters;
  CHAR_SAMPLE_IT cw_it = chars_waiting;
  CHAR_SAMPLE *sample;
  INT32 total_sample_count = 0;

  while (!cw_it.empty ()) {
    cw_it.move_to_first ();
    sample = cw_it.extract ();
    best_cluster = new CHAR_SAMPLES (sample);
    c_it.add_to_end (best_cluster);
    check_wait_list(chars_waiting, sample, best_cluster);
  }

  for (c_it.mark_cycle_pt (); !c_it.cycled_list (); c_it.forward ()) {
    c_it.data ()->assign_to_char ();
    if (tessedit_use_best_sample)
      c_it.data ()->find_best_sample ();
    else if (tessedit_mm_adapt_using_prototypes)
      c_it.data ()->build_prototype ();

    if (tessedit_cluster_debug)
      total_sample_count += c_it.data ()->n_samples ();
  }
  #ifndef SECURE_NAMES
  if (tessedit_cluster_debug)
    tprintf ("Clustering completed, %d samples in all\n", total_sample_count);
  #endif

#ifndef GRAPHICS_DISABLED
  if (tessedit_demo_adaption)
    display_cluster_prototypes(char_clusters);
#endif

}


void adapt_to_good_ems(WERD_RES *word,
                       CHAR_SAMPLES_LIST *char_clusters,
                       CHAR_SAMPLE_LIST *chars_waiting) {
  PBLOB_LIST *blobs = word->outword->blob_list ();
  PBLOB_IT blob_it(blobs);
  INT16 i;
  CHAR_SAMPLE *sample;
  CHAR_SAMPLES_IT c_it = char_clusters;
  CHAR_SAMPLE_IT cw_it = chars_waiting;
  float score;
  float best_score;
  char best_char;
  CHAR_SAMPLES *best_cluster;
  PIXROW_LIST *pixrow_list;
  PIXROW_IT pixrow_it;
  IMAGELINE *imlines;            // lines of the image
  BOX pix_box;                   // box of imlines
  // extent
  WERD copy_outword;             // copy to denorm
  BOX b_box;
  PBLOB_IT copy_blob_it;
  OUTLINE_IT copy_outline_it;
  PIXROW *pixrow = NULL;

  static INT32 word_number = 0;

#ifndef GRAPHICS_DISABLED
  WINDOW demo_win = NULL;
#endif

  INT32 resolution = page_image.get_res ();

  if (word->word->bounding_box ().height () > resolution / 3)
    return;

  word_number++;

  if (strchr (word->best_choice->string ().string (), 'm') == NULL
    && (tessedit_process_rns
    && strstr (word->best_choice->string ().string (), "rn") == NULL))
    return;

  if (tessedit_reject_ems)
    reject_all_ems(word);
  else if (tessedit_reject_suspect_ems)
    reject_suspect_ems(word);
  else {
    if (char_clusters->length () == 0) {
      #ifndef SECURE_NAMES
      if (tessedit_cluster_debug)
        tprintf ("No clusters to use for em adaption\n");
      #endif
      return;
    }

    if (!cw_it.empty ()) {
      complete_clustering(char_clusters, chars_waiting);
      print_em_stats(char_clusters, chars_waiting);
    }

    if ((!word_adaptable (word, tessedit_em_adaption_mode) ||
      word->reject_map.reject_count () != 0)
      && (strchr (word->best_choice->string ().string (), 'm') != NULL
      || (tessedit_process_rns
      && strstr (word->best_choice->string ().string (),
    "rn") != NULL))) {
      if (tessedit_process_rns
        && strstr (word->best_choice->string ().string (),
      "rn") != NULL) {
        copy_outword = *(word->outword);
        copy_blob_it.set_to_list (copy_outword.blob_list ());
        i = 0;
        while (word->best_choice->string ()[i] != '\0') {
          if (word->best_choice->string ()[i] == 'r'
          && word->best_choice->string ()[i + 1] == 'n') {
            copy_outline_it.set_to_list (copy_blob_it.data ()->
              out_list ());
            copy_outline_it.add_list_after (copy_blob_it.
              data_relative (1)->
              out_list ());
            copy_blob_it.forward ();
            delete (copy_blob_it.extract ());
            i++;
          }
          copy_blob_it.forward ();
          i++;
        }
      }
      else
        copy_outword = *(word->outword);

      copy_outword.baseline_denormalise (&word->denorm);
      copy_blob_it.set_to_list (copy_outword.blob_list ());
      char_clip_word(&copy_outword, page_image, pixrow_list, imlines, pix_box);
      pixrow_it.set_to_list (pixrow_list);
      pixrow_it.move_to_first ();

                                 // For debugging only
      b_box = copy_outword.bounding_box ();
      pixrow = pixrow_it.data ();

      blob_it.move_to_first ();
      copy_blob_it.move_to_first ();
      for (i = 0;
        word->best_choice->string ()[i] != '\0';
        i++, pixrow_it.forward (), blob_it.forward (),
      copy_blob_it.forward ()) {
        if ((word->best_choice->string ()[i] == 'm'
          || (word->best_choice->string ()[i] == 'r'
          && word->best_choice->string ()[i + 1] == 'n'))
        && !word->reject_map[i].perm_rejected ()) {
          if (tessedit_cluster_debug)
            tprintf ("Sample %c to check found in %s, index %d\n",
              word->best_choice->string ()[i],
              word->best_choice->string ().string (), i);

          if (tessedit_demo_adaption)
            tprintf
              ("Sample %c to check found in %s (%d), index %d\n",
              word->best_choice->string ()[i],
              word->best_choice->string ().string (), word_number,
              i);

          if (tessedit_matrix_match) {
            BOX copy_box = copy_blob_it.data ()->bounding_box ();

            sample = clip_sample (pixrow_it.data (),
              imlines,
              pix_box,
              copy_outword.flag (W_INVERSE),
              word->best_choice->string ()[i]);

                                 //Clip failed
            if (sample == NULL) {
              tprintf
                ("Unable to clip sample from %s, index %d\n",
                word->best_choice->string ().string (), i);
              #ifndef SECURE_NAMES
              if (tessedit_cluster_debug)
                tprintf ("Sample rejected (no sample)\n");
              #endif
              word->reject_map[i].setrej_mm_reject ();
              if (word->best_choice->string ()[i] == 'r') {
                word->reject_map[i + 1].setrej_mm_reject ();
                i++;
              }
              continue;
            }
          }
          else
            sample = new CHAR_SAMPLE (blob_it.data (),
              &word->denorm,
              word->best_choice->
              string ()[i]);

          best_score = MAX_INT32;
          best_char = '\0';
          best_cluster = NULL;

          for (c_it.mark_cycle_pt ();
          !c_it.cycled_list (); c_it.forward ()) {
            if (c_it.data ()->character () != '\0') {
              score = c_it.data ()->match_score (sample);
              if (score < best_score) {
                best_cluster = c_it.data ();
                best_score = score;
                best_char = c_it.data ()->character ();
              }
            }
          }

          if (best_score > tessedit_cluster_t1) {
            #ifndef SECURE_NAMES
            if (tessedit_cluster_debug)
              tprintf ("Sample rejected (score %f)\n", best_score);
            if (tessedit_demo_adaption)
              tprintf ("Sample rejected (score %f)\n", best_score);
            #endif
            word->reject_map[i].setrej_mm_reject ();
            if (word->best_choice->string ()[i] == 'r')
              word->reject_map[i + 1].setrej_mm_reject ();
          }
          else {
            if (word->best_choice->string ()[i] == best_char) {
              #ifndef SECURE_NAMES
              if (tessedit_cluster_debug)
                tprintf ("Sample accepted (score %f)\n",
                  best_score);
              if (tessedit_demo_adaption)
                tprintf ("Sample accepted (score %f)\n",
                  best_score);
              #endif
              word->reject_map[i].setrej_mm_accept ();
              if (word->best_choice->string ()[i] == 'r')
                word->reject_map[i + 1].setrej_mm_accept ();
            }
            else {
              #ifndef SECURE_NAMES
              if (tessedit_cluster_debug)
                tprintf ("Sample rejected (char %c, score %f)\n",
                  best_char, best_score);
              if (tessedit_demo_adaption)
                tprintf ("Sample rejected (char %c, score %f)\n",
                  best_char, best_score);
              #endif
              word->reject_map[i].setrej_mm_reject ();
              if (word->best_choice->string ()[i] == 'r')
                word->reject_map[i + 1].setrej_mm_reject ();
            }
          }

          if (tessedit_demo_adaption) {
            if (strcmp (imagebasename.string (),
              tessedit_demo_file.string ()) != 0
              || word_number == tessedit_demo_word1
            || word_number == tessedit_demo_word2) {
#ifndef GRAPHICS_DISABLED
              demo_win =
                display_clip_image(&copy_outword,
                                   page_image,
                                   pixrow_list,
                                   pix_box);
#endif
              demo_word = word_number;
              best_cluster->match_score (sample);
              demo_word = 0;
            }
          }
          if (word->best_choice->string ()[i] == 'r')
            i++;                 // Skip next character
        }
      }
      delete[]imlines;           // Free array of imlines
      delete pixrow_list;
    }
  }
}


void adapt_to_good_samples(WERD_RES *word,