adaptions.cpp

一ＯＣＲ的相关资料。．希望对研究ＯＣＲ的朋友有所帮助．

                           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;
  PIXROW *pixrow = NULL;

  static INT32 word_number = 0;

#ifndef GRAPHICS_DISABLED
  WINDOW demo_win = NULL;
#endif

  INT32 resolution = page_image.get_res ();

  word_number++;

  if (tessedit_test_cluster_input)
    return;

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

  if (char_clusters->length () == 0) {
    #ifndef SECURE_NAMES
    if (tessedit_cluster_debug)
      tprintf ("No clusters to use for 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_cluster_adaption_mode)
  && word->reject_map.reject_count () != 0) || tessedit_mm_use_rejmap) {
    if (tessedit_cluster_debug) {
      tprintf ("\nChecking: \"%s\"  MAP ",
        word->best_choice->string ().string ());
      word->reject_map.print (debug_fp);
      tprintf ("\n");
    }

    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->reject_map[i].recoverable ()
      || (tessedit_mm_all_rejects && word->reject_map[i].rejected ())) {
        BOX copy_box = copy_blob_it.data ()->bounding_box ();

        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);

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

        if (sample == NULL) {    //Clip failed
          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 ();

          continue;
        }

        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 ();
        }
        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
            if (tessedit_test_adaption)
              word->reject_map[i].setrej_minimal_rej_accept ();
            else
              word->reject_map[i].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 (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;
          }
        }
      }
    }
    delete[]imlines;             // Free array of imlines
    delete pixrow_list;

    if (tessedit_cluster_debug) {
      tprintf ("\nFinal: \"%s\"  MAP ",
        word->best_choice->string ().string ());
      word->reject_map.print (debug_fp);
      tprintf ("\n");
    }
  }
}


void print_em_stats(CHAR_SAMPLES_LIST *char_clusters,
                    CHAR_SAMPLE_LIST *chars_waiting) {
  CHAR_SAMPLES_IT c_it = char_clusters;

  if (!tessedit_cluster_debug)
    return;
  #ifndef SECURE_NAMES
  tprintf ("There are %d clusters and %d samples waiting\n",
    char_clusters->length (), chars_waiting->length ());

  for (c_it.mark_cycle_pt (); !c_it.cycled_list (); c_it.forward ())
    c_it.data ()->print (debug_fp);
  #endif
  tprintf ("\n");
}


CHAR_SAMPLE *clip_sample(              //lines of the image
                         PIXROW *pixrow,
                         IMAGELINE *imlines,
                         BOX pix_box,  //box of imlines extent
                         BOOL8 white_on_black,
                         char c) {
  BOX b_box = pixrow->bounding_box ();
  float baseline_pos = 0;
  INT32 resolution = page_image.get_res ();

  if (!b_box.null_box ()) {
    ASSERT_HOST (b_box.width () < page_image.get_xsize () &&
      b_box.height () < page_image.get_ysize ());

    if (b_box.width () > resolution || b_box.height () > resolution) {
      tprintf ("clip sample: sample too big (%d x %d)\n",
        b_box.width (), b_box.height ());

      return NULL;
    }

    IMAGE *image = new (IMAGE);
    if (image->create (b_box.width (), b_box.height (), 1) == -1) {
      tprintf ("clip sample: create image failed (%d x %d)\n",
        b_box.width (), b_box.height ());

      delete image;
      return NULL;
    }

    if (!white_on_black)
      invert_image(image);  // Set background to white
    pixrow->char_clip_image (imlines, pix_box, NULL, *image, baseline_pos);
    if (white_on_black)
      invert_image(image);  //invert white on black for scaling &NN
    return new CHAR_SAMPLE (image, c);
  }
  else
    return NULL;
}


#ifndef GRAPHICS_DISABLED
void display_cluster_prototypes(CHAR_SAMPLES_LIST *char_clusters) {
  INT16 proto_number = 0;
  CHAR_SAMPLES_IT c_it = char_clusters;
  char title[WINDOWNAMESIZE];

  for (c_it.mark_cycle_pt (); !c_it.cycled_list (); c_it.forward ()) {
    proto_number++;

    #ifndef SECURE_NAMES
    tprintf ("Displaying proto number %d\n", proto_number);
    #endif

    if (c_it.data ()->prototype () != NULL) {
      sprintf (title, "Proto - %d", proto_number);
      display_image (c_it.data ()->prototype ()->make_image (),
        title, (proto_number - 1) * 400, 0, FALSE);
    }
  }
}
#endif

// *********************************************************************
// Simplistic routines to test the effect of rejecting ems and fullstops
// *********************************************************************

void reject_all_ems(WERD_RES *word) {
  INT16 i;

  for (i = 0; word->best_choice->string ()[i] != '\0'; i++) {
    if (word->best_choice->string ()[i] == 'm')
                                 // reject all ems
      word->reject_map[i].setrej_mm_reject ();
  }
}


void reject_all_fullstops(WERD_RES *word) {
  INT16 i;

  for (i = 0; word->best_choice->string ()[i] != '\0'; i++) {
    if (word->best_choice->string ()[i] == '.')
                                 // reject all fullstops
      word->reject_map[i].setrej_mm_reject ();
  }
}


void reject_suspect_ems(WERD_RES *word) {
  INT16 i;

  if (!word_adaptable (word, tessedit_cluster_adaption_mode))
  for (i = 0; word->best_choice->string ()[i] != '\0'; i++) {
    if (word->best_choice->string ()[i] == 'm' && suspect_em (word, i))
                                 // reject all ems
      word->reject_map[i].setrej_mm_reject ();
  }
}


void reject_suspect_fullstops(WERD_RES *word) {
  INT16 i;

  for (i = 0; word->best_choice->string ()[i] != '\0'; i++) {
    if (word->best_choice->string ()[i] == '.'
      && suspect_fullstop (word, i))
                                 // reject all commas
      word->reject_map[i].setrej_mm_reject ();
  }
}


BOOL8 suspect_em(WERD_RES *word, INT16 index) {
  PBLOB_LIST *blobs = word->outword->blob_list ();
  PBLOB_IT blob_it(blobs);
  INT16 j;

  for (j = 0; j < index; j++)
    blob_it.forward ();

  return (blob_it.data ()->out_list ()->length () != 1);
}


BOOL8 suspect_fullstop(WERD_RES *word, INT16 i) {
  float aspect_ratio;
  PBLOB_LIST *blobs = word->outword->blob_list ();
  PBLOB_IT blob_it(blobs);
  INT16 j;
  BOX box;
  INT16 width;
  INT16 height;

  for (j = 0; j < i; j++)
    blob_it.forward ();

  box = blob_it.data ()->bounding_box ();

  width = box.width ();
  height = box.height ();

  aspect_ratio = ((width > height) ? ((float) width) / height :
  ((float) height) / width);

  return (aspect_ratio > tessed_fullstop_aspect_ratio);
}