chopper.cpp

一ＯＣＲ的相关资料。．希望对研究ＯＣＲ的朋友有所帮助．

    }
    if (!found_em[1] &&
      ((seam->split2 == NULL) ||
    is_split_outline (outline, seam->split2))) {
      found_em[1] = TRUE;
    }
    if (!found_em[2] &&
      ((seam->split3 == NULL) ||
    is_split_outline (outline, seam->split3))) {
      found_em[2] = TRUE;
    }
    last_outline = outline;
  }

  if (!found_em[0] || !found_em[1] || !found_em[2])
    return (FALSE);
  else
    return (TRUE);
}


/**********************************************************************
 * chop_word_main
 *
 * Classify the blobs in this word and permute the results.  Find the
 * worst blob in the word and chop it up.  Continue this process until
 * a good answer has been found or all the blobs have been chopped up
 * enough.  Return the word level ratings.
 **********************************************************************/
CHOICES_LIST chop_word_main(register TWERD *word,
                            int fx,
                            A_CHOICE *best_choice,
                            A_CHOICE *raw_choice,
                            BOOL8 tester,
                            BOOL8 trainer) {
  TBLOB *pblob;
  TBLOB *blob;
  CHOICES_LIST char_choices;
  int index;
  int did_chopping;
  float rating_limit = 1000.0;
  STATE state;
  SEAMS seam_list = NULL;
  CHOICES match_result;
  MATRIX ratings = NULL;
  DANGERR fixpt;                 /*dangerous ambig */
  INT32 state_count;             //no of states
  INT32 bit_count;               //no of bits
  static STATE best_state;
  static STATE chop_states[64];  //in between states

  state_count = 0;
  set_null_choice(best_choice);
  set_null_choice(raw_choice);

  char_choices = new_choice_list ();

  did_chopping = 0;
  for (blob = word->blobs, pblob = NULL, index = 0; blob != NULL;
  blob = blob->next, index++) {
    match_result =
      (CHOICES) classify_blob (pblob, blob, blob->next, NULL, fx,
      "chop_word:", Green, &chop_states[0],
      &best_state, matcher_pass, index);
    char_choices = array_push (char_choices, match_result);
    pblob = blob;
  }
  bit_count = index - 1;
  permute_characters(char_choices, rating_limit, best_choice, raw_choice);

  set_n_ones (&state, array_count (char_choices) - 1);
  if (matcher_fp != NULL) {
    if (matcher_pass == 0) {
      bits_in_states = bit_count;
      chop_states[state_count] = state;
    }
    state_count++;
  }

  if (!AcceptableChoice (char_choices, best_choice, raw_choice, &fixpt)
    || (tester || trainer)
  && strcmp (word->correct, class_string (best_choice))) {
    did_chopping = 1;
    if (first_pass)
      words_chopped1++;
    else
      words_chopped2++;

    seam_list = start_seam_list (word->blobs);

    if (chop_enable)
      improve_by_chopping(word,
                          &char_choices,
                          fx,
                          &state,
                          best_choice,
                          raw_choice,
                          &seam_list,
                          &fixpt,
                          chop_states,
                          &state_count,
                          &best_state,
                          matcher_pass);

    if (chop_debug)
      print_seams ("Final seam list:", seam_list);

    if (enable_assoc &&
      !AcceptableChoice (char_choices, best_choice, raw_choice, NULL)
      || (tester || trainer)
    && strcmp (word->correct, class_string (best_choice))) {
      ratings = word_associator (word->blobs, seam_list, &state, fx,
        best_choice, raw_choice, word->correct,
        /*0, */ &fixpt,
        &best_state, matcher_pass);
    }
    bits_in_states = bit_count + state_count - 1;

  }
  if (ratings != NULL)
    free_matrix(ratings);
  if (did_chopping || tester || trainer)
    char_choices = rebuild_current_state (word->blobs, seam_list, &state,
      char_choices, fx);
  if (seam_list != NULL)
    free_seam_list(seam_list);
  if (matcher_fp != NULL) {
    best_state = state;
  }
  FilterWordChoices();
  return char_choices;
}


/**********************************************************************
 * improve_by_chopping
 *
 * Start with the current word of blobs and its classification.  Find
 * the worst blobs and try to divide them up to improve the ratings.
 * As long as ratings are produced by the new blob splitting.  When
 * all the splitting has been accomplished all the ratings memory is
 * reclaimed.
 **********************************************************************/
void improve_by_chopping(register TWERD *word,
                         CHOICES_LIST *char_choices,
                         int fx,
                         STATE *best_state,
                         A_CHOICE *best_choice,
                         A_CHOICE *raw_choice,
                         SEAMS *seam_list,
                         DANGERR *fixpt,
                         STATE *chop_states,
                         INT32 *state_count,
                         STATE *correct_state,
                         INT32 pass) {
  INT32 blob_number;
  INT32 index;                   //to states
  CHOICES_LIST choices = *char_choices;
  float old_best;
  int fixpt_valid = 1;
  static INT32 old_count;        //from pass1

  do {
                                 /* Improvement loop */
    if (!fixpt_valid)
      fixpt->index = -1;
    old_best = class_probability (best_choice);
    choices = improve_one_blob (word, *char_choices, fx,
      &blob_number, seam_list, fixpt,
      chop_states + *state_count, correct_state,
      pass);
    if (choices != NULL) {
      LogNewSplit(blob_number);
      permute_characters (choices,
        class_probability (best_choice),
        best_choice, raw_choice);
      *char_choices = choices;

      if (old_best > class_probability (best_choice)) {
        set_n_ones (best_state, array_count (*char_choices) - 1);
        fixpt_valid = 1;
      }
      else {
        insert_new_chunk (best_state, blob_number,
          array_count (*char_choices) - 2);
        fixpt_valid = 0;
      }
      if (*state_count > 0) {
        if (pass == 0) {
          for (index = 0; index < *state_count; index++)
            insert_new_chunk (&chop_states[index], blob_number,
              array_count (*char_choices) - 2);
          set_n_ones (&chop_states[index],
            array_count (*char_choices) - 1);
        }
        (*state_count)++;
      }

      if (chop_debug)
        print_state ("best state = ",
          best_state, count_blobs (word->blobs) - 1);
      if (first_pass)
        chops_performed1++;
      else
        chops_performed2++;

    }
  }
  while (choices &&
    !AcceptableChoice (*char_choices, best_choice, raw_choice, fixpt) &&
    !blob_skip && array_count (*char_choices) < MAX_NUM_CHUNKS);
  if (pass == 0)
    old_count = *state_count;
  else {
    if (old_count != *state_count)
      fprintf (matcher_fp,
        "Mis-matched state counts, " INT32FORMAT " pass1, "
        INT32FORMAT " pass2\n", old_count, *state_count);
  }
  if (!fixpt_valid)
    fixpt->index = -1;
}


/**********************************************************************
 * select_blob_to_split
 *
 * These are the results of the last classification.  Find a likely
 * place to apply splits.
 **********************************************************************/
INT16 select_blob_to_split(CHOICES_LIST char_choices, float rating_ceiling) {
  CHOICES this_choice;
  int x;
  float worst = -MAX_FLOAT32;
  int worst_index = -1;

  if (chop_debug)
    if (rating_ceiling < MAX_FLOAT32)
      cprintf ("rating_ceiling = %8.4f\n", rating_ceiling);
  else
    cprintf ("rating_ceiling = No Limit\n");

  for_each_choice(char_choices, x) {
    this_choice = (CHOICES) array_value (char_choices, x);
    if (this_choice == NIL) {
      return (x);
    }
    else {
      if (best_probability (this_choice) > worst &&
        best_probability (this_choice) < rating_ceiling &&
      best_certainty (this_choice) < tessedit_certainty_threshold) {
        worst_index = x;
        worst = best_probability (this_choice);
      }
    }
  }

  if (chop_debug)
    cprintf ("blob_number = %4d\n", worst_index);

  return (worst_index);
}


/**********************************************************************
 * start_seam_list
 *
 * Initialize a list of seams that match the original number of blobs
 * present in the starting segmentation.  Each of the seams created
 * by this routine have location information only.
 **********************************************************************/
SEAMS start_seam_list(TBLOB *blobs) {
  TBLOB *blob;
  SEAMS seam_list;
  TPOINT topleft;
  TPOINT botright;
  int location;
  /* Seam slot per char */
  seam_list = new_seam_list ();

  for (blob = blobs; blob->next != NULL; blob = blob->next) {

    blob_bounding_box(blob, &topleft, &botright);
    location = botright.x;
    blob_bounding_box (blob->next, &topleft, &botright);
    location += topleft.x;
    location /= 2;

    seam_list = add_seam (seam_list,
      new_seam (0.0, location, NULL, NULL, NULL));
  }

  return (seam_list);
}


/**********************************************************************
 * total_containment
 *
 * Check to see if one of these outlines is totally contained within
 * the bounding box of the other.
 **********************************************************************/
INT16 total_containment(TBLOB *blob1, TBLOB *blob2) {
  TPOINT topleft1;
  TPOINT botright1;
  TPOINT topleft2;
  TPOINT botright2;

  blob_bounding_box(blob1, &topleft1, &botright1);
  blob_bounding_box(blob2, &topleft2, &botright2);

  return (bounds_inside (topleft1, botright1, topleft2, botright2) ||
    bounds_inside (topleft2, botright2, topleft1, botright1));
}


/**********************************************************************
 * word_associator
 *
 * Reassociate and classify the blobs in a word.  Continue this process
 * until a good answer is found or all the possibilities have been tried.
 **********************************************************************/
MATRIX word_associator(TBLOB *blobs,
                       SEAMS seams,
                       STATE *state,
                       int fxid,
                       A_CHOICE *best_choice,
                       A_CHOICE *raw_choice,
                       char *correct,
                       DANGERR *fixpt,
                       STATE *best_state,
                       INT32 pass) {
  CHUNKS_RECORD chunks_record;
  BLOB_WEIGHTS blob_weights;
  int x;
  int num_chunks;
  A_CHOICE *this_choice;

  num_chunks = array_count (seams) + 1;

  chunks_record.chunks = blobs;
  chunks_record.splits = seams;
  chunks_record.ratings = record_piece_ratings (blobs);
  chunks_record.char_widths = blobs_widths (blobs);
  chunks_record.chunk_widths = blobs_widths (blobs);
  chunks_record.fx = fxid;
  /* Save chunk weights */
  for (x = 0; x < num_chunks; x++) {
    this_choice =
      (A_CHOICE *) first (matrix_get (chunks_record.ratings, x, x));
    blob_weights[x] = -(INT16) (10 * class_probability (this_choice) /
      class_certainty (this_choice));
  }
  chunks_record.weights = blob_weights;

  if (chop_debug)
    print_matrix (chunks_record.ratings);
  best_first_search(&chunks_record,
                    best_choice,
                    raw_choice,
                    state,
                    fixpt,
                    best_state,
                    pass);

  free_widths (chunks_record.chunk_widths);
  free_widths (chunks_record.char_widths);
  return chunks_record.ratings;
}