// Remove outlines that are a tiny fraction in either width or height
// of the word height.
void Textord::clean_small_noise_from_words(ROW *row) {
WERD_IT word_it(row->word_list());
for (word_it.mark_cycle_pt(); !word_it.cycled_list(); word_it.forward()) {
WERD* word = word_it.data();
int min_size = static_cast<int>(
textord_noise_hfract * word->bounding_box().height() + 0.5);
C_BLOB_IT blob_it(word->cblob_list());
for (blob_it.mark_cycle_pt(); !blob_it.cycled_list(); blob_it.forward()) {
C_BLOB* blob = blob_it.data();
C_OUTLINE_IT out_it(blob->out_list());
for (out_it.mark_cycle_pt(); !out_it.cycled_list(); out_it.forward()) {
C_OUTLINE* outline = out_it.data();
outline->RemoveSmallRecursive(min_size, &out_it);
}
if (blob->out_list()->empty()) {
delete blob_it.extract();
}
}
if (word->cblob_list()->empty()) {
if (!word_it.at_last()) {
// The next word is no longer a fuzzy non space if it was before,
// since the word before is about to be deleted.
WERD* next_word = word_it.data_relative(1);
if (next_word->flag(W_FUZZY_NON)) {
next_word->set_flag(W_FUZZY_NON, false);
}
}
delete word_it.extract();
}
}
}
WERD *add_repeated_word( //move repeated word
WERD_IT *rep_it, //repeated words
int16_t &rep_left, //left edge of word
int16_t &prev_chop_coord, //previous word end
uint8_t &blanks, //no of blanks
float pitch, //char cell size
WERD_IT *word_it //list of words
) {
WERD *word; //word to move
int16_t new_blanks; //extra blanks
if (rep_left > prev_chop_coord) {
new_blanks = (uint8_t) floor ((rep_left - prev_chop_coord) / pitch + 0.5);
blanks += new_blanks;
}
word = rep_it->extract ();
prev_chop_coord = word->bounding_box ().right ();
word_it->add_after_then_move (word);
word->set_blanks (blanks);
rep_it->forward ();
if (rep_it->empty ())
rep_left = INT16_MAX;
else
rep_left = rep_it->data ()->bounding_box ().left ();
blanks = 0;
return word;
}
PAGE_RES_IT* make_pseudo_word(PAGE_RES* page_res, const TBOX& selection_box) {
PAGE_RES_IT pr_it(page_res);
C_BLOB_LIST new_blobs; // list of gathered blobs
C_BLOB_IT new_blob_it = &new_blobs; // iterator
for (WERD_RES* word_res = pr_it.word(); word_res != NULL;
word_res = pr_it.forward()) {
WERD* word = word_res->word;
if (word->bounding_box().overlap(selection_box)) {
C_BLOB_IT blob_it(word->cblob_list());
for (blob_it.mark_cycle_pt();
!blob_it.cycled_list(); blob_it.forward()) {
C_BLOB* blob = blob_it.data();
if (blob->bounding_box().overlap(selection_box)) {
new_blob_it.add_after_then_move(C_BLOB::deep_copy(blob));
}
}
if (!new_blobs.empty()) {
WERD* pseudo_word = new WERD(&new_blobs, 1, NULL);
word_res = pr_it.InsertSimpleCloneWord(*word_res, pseudo_word);
PAGE_RES_IT* it = new PAGE_RES_IT(page_res);
while (it->word() != word_res && it->word() != NULL) it->forward();
ASSERT_HOST(it->word() == word_res);
return it;
}
}
}
return NULL;
}
void ROW::recalc_bounding_box() { //recalculate BB
WERD *word; //current word
WERD_IT it = &words; //words of ROW
inT16 left; //of word
inT16 prev_left; //old left
if (!it.empty ()) {
word = it.data ();
prev_left = word->bounding_box ().left ();
it.forward ();
while (!it.at_first ()) {
word = it.data ();
left = word->bounding_box ().left ();
if (left < prev_left) {
it.move_to_first ();
//words in BB order
it.sort (word_comparator);
break;
}
prev_left = left;
it.forward ();
}
}
for (it.mark_cycle_pt (); !it.cycled_list (); it.forward ()) {
word = it.data ();
if (it.at_first ())
word->set_flag (W_BOL, TRUE);
else
//not start of line
word->set_flag (W_BOL, FALSE);
if (it.at_last ())
word->set_flag (W_EOL, TRUE);
else
//not end of line
word->set_flag (W_EOL, FALSE);
//extend BB as reqd
bound_box += word->bounding_box ();
}
}
/**
* @name process_selected_words()
*
* Walk the current block list applying the specified word processor function
* to each word that overlaps the selection_box.
*/
void Tesseract::process_selected_words(
PAGE_RES* page_res, // blocks to check
TBOX & selection_box,
BOOL8(tesseract::Tesseract::*word_processor)(PAGE_RES_IT* pr_it)) {
for (PAGE_RES_IT page_res_it(page_res); page_res_it.word() != NULL;
page_res_it.forward()) {
WERD* word = page_res_it.word()->word;
if (word->bounding_box().overlap(selection_box)) {
if (!(this->*word_processor)(&page_res_it))
return;
}
}
}
QString printWORD(WERD_RES *wordres,int height)
{
WERD *word = wordres->word;
QString aux;
BITS16 flags; //flags about word
flags.set_bit(W_SEGMENTED,word->flag(W_SEGMENTED));
flags.set_bit(W_ITALIC,word->flag(W_ITALIC));
flags.set_bit(W_BOL,word->flag(W_BOL));
flags.set_bit(W_EOL,word->flag(W_EOL));
flags.set_bit(W_NORMALIZED,word->flag(W_NORMALIZED));
flags.set_bit(W_POLYGON,word->flag(W_POLYGON));
flags.set_bit(W_LINEARC,word->flag(W_LINEARC));
flags.set_bit(W_DONT_CHOP,word->flag(W_DONT_CHOP));
flags.set_bit(W_REP_CHAR,word->flag(W_REP_CHAR));
flags.set_bit(W_FUZZY_SP,word->flag(W_FUZZY_SP));
flags.set_bit(W_FUZZY_NON,word->flag(W_FUZZY_NON));
aux.append(QString("Blanks= %1\n").arg(word->space()));
aux.append(printTBOX(word->bounding_box(),height,true));
aux.append(QString("Correct= %1\n")
.arg(wordres->best_choice->unichar_string().string()));
aux.append(QString("Flags = %1 = 0%2\n").arg(flags.val)
.arg(flags.val,0,8));
aux.append (QString(" W_SEGMENTED = %1\n")
.arg(word->flag(W_SEGMENTED) ? "TRUE" : "FALSE "));
aux.append (QString(" W_ITALIC = %1\n")
.arg(word->flag(W_ITALIC) ? "TRUE" : "FALSE "));
aux.append (QString(" W_BOL = %1\n")
.arg(word->flag(W_BOL) ? "TRUE" : "FALSE "));
aux.append (QString(" W_EOL = %1\n")
.arg(word->flag(W_EOL) ? "TRUE" : "FALSE "));
aux.append (QString(" W_NORMALIZED = %1\n")
.arg(word->flag(W_NORMALIZED) ? "TRUE" : "FALSE "));
aux.append (QString(" W_POLYGON = %1\n")
.arg(word->flag(W_POLYGON) ? "TRUE" : "FALSE "));
aux.append (QString(" W_LINEARC = %1\n")
.arg(word->flag(W_LINEARC) ? "TRUE" : "FALSE "));
aux.append (QString(" W_DONT_CHOP = %1\n")
.arg(word->flag(W_DONT_CHOP) ? "TRUE" : "FALSE "));
aux.append (QString(" W_REP_CHAR = %1\n")
.arg(word->flag(W_REP_CHAR) ? "TRUE" : "FALSE "));
aux.append (QString(" W_FUZZY_SP = %1\n")
.arg(word->flag(W_FUZZY_SP) ? "TRUE" : "FALSE "));
aux.append (QString(" W_FUZZY_NON = %1\n")
.arg(word->flag(W_FUZZY_NON) ? "TRUE" : "FALSE "));
aux.append(QString("Rejected cblob count = %1")
.arg(word->rej_cblob_list()->length()));
return aux;
}
/**
* Returns the baseline of the current object at the given level.
* The baseline is the line that passes through (x1, y1) and (x2, y2).
* WARNING: with vertical text, baselines may be vertical!
*/
bool PageIterator::Baseline(PageIteratorLevel level,
int* x1, int* y1, int* x2, int* y2) const {
if (it_->word() == NULL) return false; // Already at the end!
ROW* row = it_->row()->row;
WERD* word = it_->word()->word;
TBOX box = (level == RIL_WORD || level == RIL_SYMBOL)
? word->bounding_box()
: row->bounding_box();
int left = box.left();
ICOORD startpt(left, static_cast<inT16>(row->base_line(left) + 0.5));
int right = box.right();
ICOORD endpt(right, static_cast<inT16>(row->base_line(right) + 0.5));
// Rotate to image coordinates and convert to global image coords.
startpt.rotate(it_->block()->block->re_rotation());
endpt.rotate(it_->block()->block->re_rotation());
*x1 = startpt.x() / scale_ + rect_left_;
*y1 = (rect_height_ - startpt.y()) / scale_ + rect_top_;
*x2 = endpt.x() / scale_ + rect_left_;
*y2 = (rect_height_ - endpt.y()) / scale_ + rect_top_;
return true;
}
/**
* word_display() Word Processor
*
* Display a word according to its display modes
*/
BOOL8 Tesseract::word_display(BLOCK* block, ROW* row, WERD_RES* word_res) {
WERD* word = word_res->word;
TBOX word_bb; // word bounding box
int word_height; // ht of word BB
BOOL8 displayed_something = FALSE;
float shift; // from bot left
C_BLOB_IT c_it; // cblob iterator
if (color_mode != CM_RAINBOW && word_res->box_word != NULL) {
BoxWord* box_word = word_res->box_word;
int length = box_word->length();
if (word_res->fontinfo == NULL) return false;
const FontInfo& font_info = *word_res->fontinfo;
for (int i = 0; i < length; ++i) {
ScrollView::Color color = ScrollView::GREEN;
switch (color_mode) {
case CM_SUBSCRIPT:
if (box_word->BlobPosition(i) == SP_SUBSCRIPT)
color = ScrollView::RED;
break;
case CM_SUPERSCRIPT:
if (box_word->BlobPosition(i) == SP_SUPERSCRIPT)
color = ScrollView::RED;
break;
case CM_ITALIC:
if (font_info.is_italic())
color = ScrollView::RED;
break;
case CM_BOLD:
if (font_info.is_bold())
color = ScrollView::RED;
break;
case CM_FIXEDPITCH:
if (font_info.is_fixed_pitch())
color = ScrollView::RED;
break;
case CM_SERIF:
if (font_info.is_serif())
color = ScrollView::RED;
break;
case CM_SMALLCAPS:
if (word_res->small_caps)
color = ScrollView::RED;
break;
case CM_DROPCAPS:
if (box_word->BlobPosition(i) == SP_DROPCAP)
color = ScrollView::RED;
break;
// TODO(rays) underline is currently completely unsupported.
case CM_UNDERLINE:
default:
break;
}
image_win->Pen(color);
TBOX box = box_word->BlobBox(i);
image_win->Rectangle(box.left(), box.bottom(), box.right(), box.top());
}
return true;
}
/*
Note the double coercions of(COLOUR)((inT32)editor_image_word_bb_color)
etc. are to keep the compiler happy.
*/
// display bounding box
if (word->display_flag(DF_BOX)) {
word->bounding_box().plot(image_win,
(ScrollView::Color)((inT32)
editor_image_word_bb_color),
(ScrollView::Color)((inT32)
editor_image_word_bb_color));
ScrollView::Color c = (ScrollView::Color)
((inT32) editor_image_blob_bb_color);
image_win->Pen(c);
c_it.set_to_list(word->cblob_list());
for (c_it.mark_cycle_pt(); !c_it.cycled_list(); c_it.forward())
c_it.data()->bounding_box().plot(image_win);
displayed_something = TRUE;
}
// display edge steps
if (word->display_flag(DF_EDGE_STEP)) { // edgesteps available
word->plot(image_win); // rainbow colors
displayed_something = TRUE;
}
// display poly approx
if (word->display_flag(DF_POLYGONAL)) {
// need to convert
TWERD* tword = TWERD::PolygonalCopy(word);
tword->plot(image_win);
delete tword;
displayed_something = TRUE;
}
// Display correct text and blamer information.
STRING text;
STRING blame;
if (word->display_flag(DF_TEXT) && word->text() != NULL) {
text = word->text();
}
if (word->display_flag(DF_BLAMER) &&
!(word_res->blamer_bundle != NULL &&
word_res->blamer_bundle->incorrect_result_reason == IRR_CORRECT)) {
text = "";
const BlamerBundle *blamer_bundle = word_res->blamer_bundle;
if (blamer_bundle == NULL) {
text += "NULL";
} else {
for (int i = 0; i < blamer_bundle->truth_text.length(); ++i) {
text += blamer_bundle->truth_text[i];
}
}
text += " -> ";
STRING best_choice_str;
if (word_res->best_choice == NULL) {
best_choice_str = "NULL";
} else {
word_res->best_choice->string_and_lengths(&best_choice_str, NULL);
}
text += best_choice_str;
IncorrectResultReason reason = (blamer_bundle == NULL) ?
IRR_PAGE_LAYOUT : blamer_bundle->incorrect_result_reason;
ASSERT_HOST(reason < IRR_NUM_REASONS)
blame += " [";
blame += BlamerBundle::IncorrectReasonName(reason);
blame += "]";
}
if (text.length() > 0) {
word_bb = word->bounding_box();
image_win->Pen(ScrollView::RED);
word_height = word_bb.height();
int text_height = 0.50 * word_height;
if (text_height > 20) text_height = 20;
image_win->TextAttributes("Arial", text_height, false, false, false);
shift = (word_height < word_bb.width()) ? 0.25 * word_height : 0.0f;
image_win->Text(word_bb.left() + shift,
word_bb.bottom() + 0.25 * word_height, text.string());
if (blame.length() > 0) {
image_win->Text(word_bb.left() + shift,
word_bb.bottom() + 0.25 * word_height - text_height,
blame.string());
}
displayed_something = TRUE;
}
if (!displayed_something) // display BBox anyway
word->bounding_box().plot(image_win,
(ScrollView::Color)((inT32) editor_image_word_bb_color),
(ScrollView::Color)((inT32)
editor_image_word_bb_color));
return TRUE;
}
/*************************************************************************
* tidy_up()
* - report >1 block
* - sort the words in each row.
* - report any rows with no labelled words.
* - report any remaining unlabelled words
* - report total labelled words
*
*************************************************************************/
void tidy_up( //
BLOCK_LIST *block_list, //real blocks
INT16 &ok_char_count,
INT16 &ok_row_count,
INT16 &unlabelled_words,
INT16 *tgt_char_counts,
INT16 &rebalance_count,
char &min_char,
INT16 &min_samples,
INT16 &final_labelled_blob_count) {
BLOCK_IT block_it(block_list);
ROW_IT row_it;
ROW *row;
WERD_IT word_it;
WERD *word;
WERD *duplicate_word;
INT16 block_idx = 0;
INT16 row_idx;
INT16 all_row_idx = 0;
BOOL8 row_ok;
BOOL8 rebalance_needed = FALSE;
//No. of unique labelled samples
INT16 labelled_char_counts[128];
INT16 i;
char ch;
char prev_ch = '\0';
BOOL8 at_dupe_of_prev_word;
ROW *prev_row = NULL;
INT16 left;
INT16 prev_left = -1;
for (i = 0; i < 128; i++)
labelled_char_counts[i] = 0;
ok_char_count = 0;
ok_row_count = 0;
unlabelled_words = 0;
if ((applybox_debug > 4) && (block_it.length () != 1))
tprintf ("APPLY_BOXES: More than one block??\n");
for (block_it.mark_cycle_pt ();
!block_it.cycled_list (); block_it.forward ()) {
block_idx++;
row_idx = 0;
row_ok = FALSE;
row_it.set_to_list (block_it.data ()->row_list ());
for (row_it.mark_cycle_pt (); !row_it.cycled_list (); row_it.forward ()) {
row_idx++;
all_row_idx++;
row = row_it.data ();
word_it.set_to_list (row->word_list ());
word_it.sort (word_comparator);
for (word_it.mark_cycle_pt ();
!word_it.cycled_list (); word_it.forward ()) {
word = word_it.data ();
if (strlen (word->text ()) == 0) {
unlabelled_words++;
if (applybox_debug > 4) {
tprintf
("APPLY_BOXES: Unlabelled word blk:%d row:%d allrows:%d\n",
block_idx, row_idx, all_row_idx);
}
}
else {
if (word->gblob_list ()->length () != 1)
tprintf
("APPLY_BOXES: FATALITY - MULTIBLOB Labelled word blk:%d row:%d allrows:%d\n",
block_idx, row_idx, all_row_idx);
ok_char_count++;
labelled_char_counts[*word->text ()]++;
row_ok = TRUE;
}
}
if ((applybox_debug > 4) && (!row_ok)) {
tprintf
("APPLY_BOXES: Row with no labelled words blk:%d row:%d allrows:%d\n",
block_idx, row_idx, all_row_idx);
}
else
ok_row_count++;
}
}
min_samples = 9999;
for (i = 0; i < 128; i++) {
if (tgt_char_counts[i] > labelled_char_counts[i]) {
if (labelled_char_counts[i] <= 1) {
tprintf
("APPLY_BOXES: FATALITY - %d labelled samples of \"%c\" - target is %d\n",
labelled_char_counts[i], (char) i, tgt_char_counts[i]);
}
else {
rebalance_needed = TRUE;
if (applybox_debug > 0)
tprintf
("APPLY_BOXES: REBALANCE REQD \"%c\" - target of %d from %d labelled samples\n",
(char) i, tgt_char_counts[i], labelled_char_counts[i]);
}
}
if ((min_samples > labelled_char_counts[i]) && (tgt_char_counts[i] > 0)) {
min_samples = labelled_char_counts[i];
min_char = (char) i;
}
}
while (applybox_rebalance && rebalance_needed) {
block_it.set_to_list (block_list);
for (block_it.mark_cycle_pt ();
!block_it.cycled_list (); block_it.forward ()) {
row_it.set_to_list (block_it.data ()->row_list ());
for (row_it.mark_cycle_pt ();
!row_it.cycled_list (); row_it.forward ()) {
row = row_it.data ();
word_it.set_to_list (row->word_list ());
for (word_it.mark_cycle_pt ();
!word_it.cycled_list (); word_it.forward ()) {
word = word_it.data ();
left = word->bounding_box ().left ();
ch = *word->text ();
at_dupe_of_prev_word = ((row == prev_row) &&
(left = prev_left) &&
(ch == prev_ch));
if ((ch != '\0') &&
(labelled_char_counts[ch] > 1) &&
(tgt_char_counts[ch] > labelled_char_counts[ch]) &&
(!at_dupe_of_prev_word)) {
/* Duplicate the word to rebalance the labelled samples */
if (applybox_debug > 9) {
tprintf ("Duping \"%c\" from ", ch);
word->bounding_box ().print ();
}
duplicate_word = new WERD;
*duplicate_word = *word;
word_it.add_after_then_move (duplicate_word);
rebalance_count++;
labelled_char_counts[ch]++;
}
prev_row = row;
prev_left = left;
prev_ch = ch;
}
}
}
rebalance_needed = FALSE;
for (i = 0; i < 128; i++) {
if ((tgt_char_counts[i] > labelled_char_counts[i]) &&
(labelled_char_counts[i] > 1)) {
rebalance_needed = TRUE;
break;
}
}
}
/* Now final check - count labelled blobs */
final_labelled_blob_count = 0;
block_it.set_to_list (block_list);
for (block_it.mark_cycle_pt ();
!block_it.cycled_list (); block_it.forward ()) {
row_it.set_to_list (block_it.data ()->row_list ());
for (row_it.mark_cycle_pt (); !row_it.cycled_list (); row_it.forward ()) {
row = row_it.data ();
word_it.set_to_list (row->word_list ());
word_it.sort (word_comparator);
for (word_it.mark_cycle_pt ();
!word_it.cycled_list (); word_it.forward ()) {
word = word_it.data ();
if ((strlen (word->text ()) == 1) &&
(word->gblob_list ()->length () == 1))
final_labelled_blob_count++;
}
}
}
}
INT16 resegment_box( //
ROW *row,
BOX box,
char *ch,
INT16 block_id,
INT16 row_id,
INT16 boxfile_lineno,
INT16 boxfile_charno) {
WERD_IT word_it;
WERD *word;
WERD *new_word = NULL;
BOOL8 polyg = false;
PBLOB_IT blob_it;
PBLOB_IT new_blob_it;
PBLOB *blob;
PBLOB *new_blob;
OUTLINE_IT outline_it;
OUTLINE_LIST dummy; // Just to initialize new_outline_it.
OUTLINE_IT new_outline_it = &dummy;
OUTLINE *outline;
BOX new_word_box;
float word_x_centre;
float baseline;
INT16 error_count = 0; //number of chars lost
word_it.set_to_list (row->word_list ());
for (word_it.mark_cycle_pt (); !word_it.cycled_list (); word_it.forward ()) {
word = word_it.data ();
polyg = word->flag (W_POLYGON);
if (word->bounding_box ().overlap (box)) {
blob_it.set_to_list (word->gblob_list ());
for (blob_it.mark_cycle_pt ();
!blob_it.cycled_list (); blob_it.forward ()) {
blob = blob_it.data ();
if (gblob_bounding_box (blob, polyg).overlap (box)) {
outline_it.set_to_list (gblob_out_list (blob, polyg));
for (outline_it.mark_cycle_pt ();
!outline_it.cycled_list (); outline_it.forward ()) {
outline = outline_it.data ();
if (goutline_bounding_box (outline, polyg).
major_overlap (box)) {
if (strlen (word->text ()) > 0) {
if (error_count == 0) {
error_count = 1;
if (applybox_debug > 4)
report_failed_box (boxfile_lineno,
boxfile_charno,
box, ch,
"FAILURE! box overlaps blob in labelled word");
}
if (applybox_debug > 4)
tprintf
("APPLY_BOXES: ALSO ignoring corrupted char blk:%d row:%d \"%s\"\n",
block_id, row_id,
word_it.data ()->text ());
word_it.data ()->set_text ("");
//UN label it
error_count++;
}
if (error_count == 0) {
if (new_word == NULL) {
/* Make a new word with a single blob */
new_word = word->shallow_copy ();
new_word->set_text (ch);
if (polyg)
new_blob = new PBLOB;
else
new_blob = (PBLOB *) new C_BLOB;
new_blob_it.set_to_list (new_word->
gblob_list ());
new_blob_it.add_to_end (new_blob);
new_outline_it.
set_to_list (gblob_out_list
(new_blob, polyg));
}
new_outline_it.add_to_end (outline_it.
extract ());
//move blob
}
}
}
//no outlines in blob
if (outline_it.empty ())
//so delete blob
delete blob_it.extract ();
}
}
if (blob_it.empty ()) //no blobs in word
//so delete word
delete word_it.extract ();
}
}
if (error_count > 0)
return error_count;
if (new_word != NULL) {
gblob_sort_list (new_word->gblob_list (), polyg);
word_it.add_to_end (new_word);
new_word_box = new_word->bounding_box ();
word_x_centre = (new_word_box.left () + new_word_box.right ()) / 2.0f;
baseline = row->base_line (word_x_centre);
if (STRING (chs_caps_ht).contains (ch[0]) &&
(new_word_box.top () <
baseline + (1 + applybox_error_band) * row->x_height ())) {
report_failed_box (boxfile_lineno, boxfile_charno, box, ch,
"FAILURE! caps-ht char didn't ascend");
new_word->set_text ("");
return 1;
}
if (STRING (chs_odd_top).contains (ch[0]) &&
(new_word_box.top () <
baseline + (1 - applybox_error_band) * row->x_height ())) {
report_failed_box (boxfile_lineno, boxfile_charno, box, ch,
"FAILURE! Odd top char below xht");
new_word->set_text ("");
return 1;
}
if (STRING (chs_x_ht).contains (ch[0]) &&
((new_word_box.top () >
baseline + (1 + applybox_error_band) * row->x_height ()) ||
(new_word_box.top () <
baseline + (1 - applybox_error_band) * row->x_height ()))) {
report_failed_box (boxfile_lineno, boxfile_charno, box, ch,
"FAILURE! x-ht char didn't have top near xht");
new_word->set_text ("");
return 1;
}
if (STRING (chs_non_ambig_bl).contains (ch[0]) &&
((new_word_box.bottom () <
baseline - applybox_error_band * row->x_height ()) ||
(new_word_box.bottom () >
baseline + applybox_error_band * row->x_height ()))) {
report_failed_box (boxfile_lineno, boxfile_charno, box, ch,
"FAILURE! non ambig BL char didnt have bottom near baseline");
new_word->set_text ("");
return 1;
}
if (STRING (chs_odd_bot).contains (ch[0]) &&
(new_word_box.bottom () >
baseline + applybox_error_band * row->x_height ())) {
report_failed_box (boxfile_lineno, boxfile_charno, box, ch,
"FAILURE! Odd bottom char above baseline");
new_word->set_text ("");
return 1;
}
if (STRING (chs_desc).contains (ch[0]) &&
(new_word_box.bottom () >
baseline - applybox_error_band * row->x_height ())) {
report_failed_box (boxfile_lineno, boxfile_charno, box, ch,
"FAILURE! Descender doesn't descend");
new_word->set_text ("");
return 1;
}
return 0;
}
else {
report_failed_box (boxfile_lineno, boxfile_charno, box, ch,
"FAILURE! Couldn't find any blobs");
return 1;
}
}
ROW *find_row_of_box( //
BLOCK_LIST *block_list, //real blocks
BOX box, //from boxfile
INT16 &block_id,
INT16 &row_id_to_process) {
BLOCK_IT block_it(block_list);
BLOCK *block;
ROW_IT row_it;
ROW *row;
ROW *row_to_process = NULL;
INT16 row_id;
WERD_IT word_it;
WERD *word;
BOOL8 polyg;
PBLOB_IT blob_it;
PBLOB *blob;
OUTLINE_IT outline_it;
OUTLINE *outline;
/*
Find row to process - error if box REALLY overlaps more than one row. (I.e
it overlaps blobs in the row - not just overlaps the bounding box of the
whole row.)
*/
block_id = 0;
for (block_it.mark_cycle_pt ();
!block_it.cycled_list (); block_it.forward ()) {
block_id++;
row_id = 0;
block = block_it.data ();
if (block->bounding_box ().overlap (box)) {
row_it.set_to_list (block->row_list ());
for (row_it.mark_cycle_pt ();
!row_it.cycled_list (); row_it.forward ()) {
row_id++;
row = row_it.data ();
if (row->bounding_box ().overlap (box)) {
word_it.set_to_list (row->word_list ());
for (word_it.mark_cycle_pt ();
!word_it.cycled_list (); word_it.forward ()) {
word = word_it.data ();
polyg = word->flag (W_POLYGON);
if (word->bounding_box ().overlap (box)) {
blob_it.set_to_list (word->gblob_list ());
for (blob_it.mark_cycle_pt ();
!blob_it.cycled_list (); blob_it.forward ()) {
blob = blob_it.data ();
if (gblob_bounding_box (blob, polyg).
overlap (box)) {
outline_it.
set_to_list (gblob_out_list
(blob, polyg));
for (outline_it.mark_cycle_pt ();
!outline_it.cycled_list ();
outline_it.forward ()) {
outline = outline_it.data ();
if (goutline_bounding_box
(outline, polyg).major_overlap (box)) {
if ((row_to_process == NULL) ||
(row_to_process == row)) {
row_to_process = row;
row_id_to_process = row_id;
}
else
/* RETURN ERROR Box overlaps blobs in more than one row */
return NULL;
}
}
}
}
}
}
}
}
}
}
return row_to_process;
}
/// Consume all source blobs that strongly overlap the given box,
/// putting them into a new word, with the correct_text label.
/// Fights over which box owns which blobs are settled by
/// applying the blobs to box or next_box with the least non-overlap.
/// @return false if the box was in error, which can only be caused by
/// failing to find an overlapping blob for a box.
bool Tesseract::ResegmentWordBox(BLOCK_LIST *block_list,
const TBOX& box, const TBOX& next_box,
const char* correct_text) {
if (applybox_debug > 1) {
tprintf("\nAPPLY_BOX: in ResegmentWordBox() for %s\n", correct_text);
}
WERD* new_word = NULL;
BLOCK_IT b_it(block_list);
for (b_it.mark_cycle_pt(); !b_it.cycled_list(); b_it.forward()) {
BLOCK* block = b_it.data();
if (!box.major_overlap(block->bounding_box()))
continue;
ROW_IT r_it(block->row_list());
for (r_it.mark_cycle_pt(); !r_it.cycled_list(); r_it.forward()) {
ROW* row = r_it.data();
if (!box.major_overlap(row->bounding_box()))
continue;
WERD_IT w_it(row->word_list());
for (w_it.mark_cycle_pt(); !w_it.cycled_list(); w_it.forward()) {
WERD* word = w_it.data();
if (applybox_debug > 2) {
tprintf("Checking word:");
word->bounding_box().print();
}
if (word->text() != NULL && word->text()[0] != '\0')
continue; // Ignore words that are already done.
if (!box.major_overlap(word->bounding_box()))
continue;
C_BLOB_IT blob_it(word->cblob_list());
for (blob_it.mark_cycle_pt(); !blob_it.cycled_list();
blob_it.forward()) {
C_BLOB* blob = blob_it.data();
TBOX blob_box = blob->bounding_box();
if (!blob_box.major_overlap(box))
continue;
double current_box_miss_metric = BoxMissMetric(blob_box, box);
double next_box_miss_metric = BoxMissMetric(blob_box, next_box);
if (applybox_debug > 2) {
tprintf("Checking blob:");
blob_box.print();
tprintf("Current miss metric = %g, next = %g\n",
current_box_miss_metric, next_box_miss_metric);
}
if (current_box_miss_metric > next_box_miss_metric)
continue; // Blob is a better match for next box.
if (applybox_debug > 2) {
tprintf("Blob match: blob:");
blob_box.print();
tprintf("Matches box:");
box.print();
tprintf("With next box:");
next_box.print();
}
if (new_word == NULL) {
// Make a new word with a single blob.
new_word = word->shallow_copy();
new_word->set_text(correct_text);
w_it.add_to_end(new_word);
}
C_BLOB_IT new_blob_it(new_word->cblob_list());
new_blob_it.add_to_end(blob_it.extract());
}
}
}
}
if (new_word == NULL && applybox_debug > 0) tprintf("FAIL!\n");
return new_word != NULL;
}