private string OcrViaNOCR(Bitmap bitmap, int listViewIndex)
{
if (_ocrFixEngine == null)
comboBoxDictionaries_SelectedIndexChanged(null, null);
string line = string.Empty;
if (_nocrThreadResults != null)
line = _nocrThreadResults[listViewIndex];
if (string.IsNullOrEmpty(line))
{
var nbmpInput = new NikseBitmap(bitmap);
var matches = new List<CompareMatch>();
int minLineHeight = _binOcrLastLowercaseHeight - 3;
if (minLineHeight < 5)
minLineHeight = _nocrLastLowercaseHeight - 3;
if (minLineHeight < 5)
minLineHeight = 5;
List<ImageSplitterItem> list = NikseBitmapImageSplitter.SplitBitmapToLettersNew(nbmpInput, (int)numericUpDownNumberOfPixelsIsSpaceNOCR.Value, checkBoxRightToLeft.Checked, Configuration.Settings.VobSubOcr.TopToBottom, minLineHeight);
foreach (ImageSplitterItem item in list)
{
if (item.NikseBitmap != null)
{
item.NikseBitmap.ReplaceTransparentWith(Color.Black);
}
}
int index = 0;
bool expandSelection = false;
bool shrinkSelection = false;
var expandSelectionList = new List<ImageSplitterItem>();
while (index < list.Count)
{
ImageSplitterItem item = list[index];
if (expandSelection || shrinkSelection)
{
expandSelection = false;
if (shrinkSelection && index > 0)
{
shrinkSelection = false;
}
else if (index + 1 < list.Count && list[index + 1].NikseBitmap != null) // only allow expand to EndOfLine or space
{
index++;
expandSelectionList.Add(list[index]);
}
item = GetExpandedSelection(nbmpInput, expandSelectionList, checkBoxRightToLeft.Checked);
if (item.NikseBitmap != null)
{
item.NikseBitmap.ReplaceTransparentWith(Color.Black);
}
_vobSubOcrNOcrCharacter.Initialize(bitmap, item, _manualOcrDialogPosition, _italicCheckedLast, expandSelectionList.Count > 1, null, _lastAdditions, this);
DialogResult result = _vobSubOcrNOcrCharacter.ShowDialog(this);
_manualOcrDialogPosition = _vobSubOcrNOcrCharacter.FormPosition;
if (result == DialogResult.OK && _vobSubOcrNOcrCharacter.ShrinkSelection)
{
shrinkSelection = true;
index--;
if (expandSelectionList.Count > 0)
expandSelectionList.RemoveAt(expandSelectionList.Count - 1);
}
else if (result == DialogResult.OK && _vobSubOcrNOcrCharacter.ExpandSelection)
{
expandSelection = true;
}
else if (result == DialogResult.OK)
{
var c = _vobSubOcrNOcrCharacter.NOcrChar;
if (expandSelectionList.Count > 1)
c.ExpandCount = expandSelectionList.Count;
_nOcrDb.Add(c);
SaveNOcrWithCurrentLanguage();
string text = _vobSubOcrNOcrCharacter.NOcrChar.Text;
string name = SaveCompareItem(item.NikseBitmap, text, _vobSubOcrNOcrCharacter.IsItalic, expandSelectionList.Count);
var addition = new ImageCompareAddition(name, text, item.NikseBitmap, _vobSubOcrNOcrCharacter.IsItalic, listViewIndex);
_lastAdditions.Add(addition);
matches.Add(new CompareMatch(text, _vobSubOcrNOcrCharacter.IsItalic, expandSelectionList.Count, null));
expandSelectionList = new List<ImageSplitterItem>();
}
else if (result == DialogResult.Abort)
{
_abort = true;
}
else
{
matches.Add(new CompareMatch("*", false, 0, null));
}
_italicCheckedLast = _vobSubOcrNOcrCharacter.IsItalic;
}
else if (item.NikseBitmap == null)
{
matches.Add(new CompareMatch(item.SpecialCharacter, false, 0, null));
}
else
{
CompareMatch match = GetNOcrCompareMatchNew(item, nbmpInput, _nOcrDb, _unItalicFactor, checkBoxNOcrItalic.Checked, !checkBoxNOcrCorrect.Checked);
if (match == null)
{
_vobSubOcrNOcrCharacter.Initialize(bitmap, item, _manualOcrDialogPosition, _italicCheckedLast, false, null, _lastAdditions, this);
DialogResult result = _vobSubOcrNOcrCharacter.ShowDialog(this);
_manualOcrDialogPosition = _vobSubOcrNOcrCharacter.FormPosition;
if (result == DialogResult.OK && _vobSubOcrNOcrCharacter.ExpandSelection)
{
expandSelectionList.Add(item);
expandSelection = true;
}
else if (result == DialogResult.OK)
{
_nOcrDb.Add(_vobSubOcrNOcrCharacter.NOcrChar);
SaveNOcrWithCurrentLanguage();
string text = _vobSubOcrNOcrCharacter.NOcrChar.Text;
string name = SaveCompareItem(item.NikseBitmap, text, _vobSubOcrNOcrCharacter.IsItalic, 0);
var addition = new ImageCompareAddition(name, text, item.NikseBitmap, _vobSubOcrNOcrCharacter.IsItalic, listViewIndex);
_lastAdditions.Add(addition);
matches.Add(new CompareMatch(text, _vobSubOcrNOcrCharacter.IsItalic, 0, null));
}
else if (result == DialogResult.Abort)
{
_abort = true;
}
else
{
matches.Add(new CompareMatch("*", false, 0, null));
}
_italicCheckedLast = _vobSubOcrNOcrCharacter.IsItalic;
}
else // found image match
{
matches.Add(new CompareMatch(match.Text, match.Italic, 0, null));
if (match.ExpandCount > 0)
index += match.ExpandCount - 1;
}
}
if (_abort)
return string.Empty;
if (!expandSelection && !shrinkSelection)
index++;
if (shrinkSelection && expandSelectionList.Count < 2)
{
shrinkSelection = false;
expandSelectionList = new List<ImageSplitterItem>();
}
}
line = GetStringWithItalicTags(matches);
}
line = FixNocrHardcodedStuff(line);
//ocr fix engine
string textWithOutFixes = line;
if (_ocrFixEngine.IsDictionaryLoaded)
{
if (checkBoxAutoFixCommonErrors.Checked)
line = _ocrFixEngine.FixOcrErrors(line, listViewIndex, _lastLine, true, checkBoxGuessUnknownWords.Checked);
int correctWords;
int wordsNotFound = _ocrFixEngine.CountUnknownWordsViaDictionary(line, out correctWords);
if (wordsNotFound > 0 || correctWords == 0 || textWithOutFixes != null && textWithOutFixes.ToString().Replace("~", string.Empty).Trim().Length == 0)
{
_ocrFixEngine.AutoGuessesUsed.Clear();
_ocrFixEngine.UnknownWordsFound.Clear();
line = _ocrFixEngine.FixUnknownWordsViaGuessOrPrompt(out wordsNotFound, line, listViewIndex, bitmap, checkBoxAutoFixCommonErrors.Checked, checkBoxPromptForUnknownWords.Checked, true, checkBoxGuessUnknownWords.Checked);
}
if (_ocrFixEngine.Abort)
{
ButtonStopClick(null, null);
_ocrFixEngine.Abort = false;
return string.Empty;
}
// Log used word guesses (via word replace list)
foreach (string guess in _ocrFixEngine.AutoGuessesUsed)
listBoxLogSuggestions.Items.Add(guess);
_ocrFixEngine.AutoGuessesUsed.Clear();
// Log unkown words guess (found via spelling dictionaries)
LogUnknownWords();
if (wordsNotFound >= 3)
subtitleListView1.SetBackgroundColor(listViewIndex, Color.Red);
if (wordsNotFound == 2)
subtitleListView1.SetBackgroundColor(listViewIndex, Color.Orange);
else if (wordsNotFound == 1)
subtitleListView1.SetBackgroundColor(listViewIndex, Color.Yellow);
else if (line.Trim().Length == 0)
subtitleListView1.SetBackgroundColor(listViewIndex, Color.Orange);
else
subtitleListView1.SetBackgroundColor(listViewIndex, Color.LightGreen);
}
if (textWithOutFixes.Trim() != line.Trim())
{
_tesseractOcrAutoFixes++;
labelFixesMade.Text = string.Format(" - {0}", _tesseractOcrAutoFixes);
LogOcrFix(listViewIndex, textWithOutFixes.ToString(), line);
}
return line;
}
/// <summary>
/// Ocr via image compare
/// </summary>
private string SplitAndOcrBitmapNormal(Bitmap bitmap, int listViewIndex)
{
if (_ocrFixEngine == null)
LoadOcrFixEngine(null, LanguageString);
string threadText = null;
if (_icThreadResults != null && !string.IsNullOrEmpty(_icThreadResults[listViewIndex]))
threadText = _icThreadResults[listViewIndex];
string line = string.Empty;
if (threadText == null)
{
var matches = new List<CompareMatch>();
var parentBitmap = new NikseBitmap(bitmap);
List<ImageSplitterItem> list = NikseBitmapImageSplitter.SplitBitmapToLetters(parentBitmap, (int)numericUpDownPixelsIsSpace.Value, checkBoxRightToLeft.Checked, Configuration.Settings.VobSubOcr.TopToBottom);
int index = 0;
bool expandSelection = false;
bool shrinkSelection = false;
var expandSelectionList = new List<ImageSplitterItem>();
while (index < list.Count)
{
ImageSplitterItem item = list[index];
if (expandSelection || shrinkSelection)
{
expandSelection = false;
if (shrinkSelection && index > 0)
{
shrinkSelection = false;
}
else if (index + 1 < list.Count && list[index + 1].NikseBitmap != null) // only allow expand to EndOfLine or space
{
index++;
expandSelectionList.Add(list[index]);
}
item = GetExpandedSelection(parentBitmap, expandSelectionList, checkBoxRightToLeft.Checked);
_vobSubOcrCharacter.Initialize(bitmap, item, _manualOcrDialogPosition, _italicCheckedLast, expandSelectionList.Count > 1, null, _lastAdditions, this);
DialogResult result = _vobSubOcrCharacter.ShowDialog(this);
_manualOcrDialogPosition = _vobSubOcrCharacter.FormPosition;
if (result == DialogResult.OK && _vobSubOcrCharacter.ShrinkSelection)
{
shrinkSelection = true;
index--;
if (expandSelectionList.Count > 0)
expandSelectionList.RemoveAt(expandSelectionList.Count - 1);
}
else if (result == DialogResult.OK && _vobSubOcrCharacter.ExpandSelection)
{
expandSelection = true;
}
else if (result == DialogResult.OK)
{
string text = _vobSubOcrCharacter.ManualRecognizedCharacters;
string name = SaveCompareItem(item.NikseBitmap, text, _vobSubOcrCharacter.IsItalic, expandSelectionList.Count);
var addition = new ImageCompareAddition(name, text, item.NikseBitmap, _vobSubOcrCharacter.IsItalic, listViewIndex);
_lastAdditions.Add(addition);
matches.Add(new CompareMatch(text, _vobSubOcrCharacter.IsItalic, expandSelectionList.Count, null));
expandSelectionList = new List<ImageSplitterItem>();
}
else if (result == DialogResult.Abort)
{
_abort = true;
}
else
{
matches.Add(new CompareMatch("*", false, 0, null));
}
_italicCheckedLast = _vobSubOcrCharacter.IsItalic;
}
else if (item.NikseBitmap == null)
{
matches.Add(new CompareMatch(item.SpecialCharacter, false, 0, null));
}
else
{
CompareMatch bestGuess;
CompareMatch match = GetCompareMatch(item, parentBitmap, out bestGuess, list, index);
if (match == null)
{
_vobSubOcrCharacter.Initialize(bitmap, item, _manualOcrDialogPosition, _italicCheckedLast, false, bestGuess, _lastAdditions, this);
DialogResult result = _vobSubOcrCharacter.ShowDialog(this);
_manualOcrDialogPosition = _vobSubOcrCharacter.FormPosition;
if (result == DialogResult.OK && _vobSubOcrCharacter.ExpandSelection)
{
expandSelectionList.Add(item);
expandSelection = true;
}
else if (result == DialogResult.OK)
{
string text = _vobSubOcrCharacter.ManualRecognizedCharacters;
string name = SaveCompareItem(item.NikseBitmap, text, _vobSubOcrCharacter.IsItalic, 0);
var addition = new ImageCompareAddition(name, text, item.NikseBitmap, _vobSubOcrCharacter.IsItalic, listViewIndex);
_lastAdditions.Add(addition);
matches.Add(new CompareMatch(text, _vobSubOcrCharacter.IsItalic, 0, null));
}
else if (result == DialogResult.Abort)
{
_abort = true;
}
else
{
matches.Add(new CompareMatch("*", false, 0, null));
}
_italicCheckedLast = _vobSubOcrCharacter.IsItalic;
}
else // found image match
{
matches.Add(new CompareMatch(match.Text, match.Italic, 0, null));
if (match.ExpandCount > 0)
index += match.ExpandCount - 1;
}
}
if (_abort)
return string.Empty;
if (!expandSelection && !shrinkSelection)
index++;
if (shrinkSelection && expandSelectionList.Count < 2)
{
shrinkSelection = false;
expandSelectionList = new List<ImageSplitterItem>();
}
}
line = GetStringWithItalicTags(matches);
}
else
{
line = threadText;
}
if (checkBoxAutoFixCommonErrors.Checked && _ocrFixEngine != null)
line = _ocrFixEngine.FixOcrErrorsViaHardcodedRules(line, _lastLine, null); // TODO: Add abbreviations list
if (checkBoxRightToLeft.Checked)
line = ReverseNumberStrings(line);
//OCR fix engine
string textWithOutFixes = line;
//OCR fix engine not loaded, when no dictionary is selected
if (_ocrFixEngine != null && _ocrFixEngine.IsDictionaryLoaded)
{
if (checkBoxAutoFixCommonErrors.Checked)
line = _ocrFixEngine.FixOcrErrors(line, listViewIndex, _lastLine, true, GetAutoGuessLevel());
int correctWords;
int wordsNotFound = _ocrFixEngine.CountUnknownWordsViaDictionary(line, out correctWords);
if (wordsNotFound > 0 || correctWords == 0 || textWithOutFixes != null && string.IsNullOrWhiteSpace(textWithOutFixes.Replace("~", string.Empty)))
{
_ocrFixEngine.AutoGuessesUsed.Clear();
_ocrFixEngine.UnknownWordsFound.Clear();
line = _ocrFixEngine.FixUnknownWordsViaGuessOrPrompt(out wordsNotFound, line, listViewIndex, bitmap, checkBoxAutoFixCommonErrors.Checked, checkBoxPromptForUnknownWords.Checked, true, GetAutoGuessLevel());
}
if (_ocrFixEngine.Abort)
{
ButtonStopClick(null, null);
_ocrFixEngine.Abort = false;
return string.Empty;
}
// Log used word guesses (via word replace list)
foreach (string guess in _ocrFixEngine.AutoGuessesUsed)
listBoxLogSuggestions.Items.Add(guess);
_ocrFixEngine.AutoGuessesUsed.Clear();
// Log unkown words guess (found via spelling dictionaries)
LogUnknownWords();
ColorLineByNumberOfUnknownWords(listViewIndex, wordsNotFound, line);
}
if (textWithOutFixes.Trim() != line.Trim())
{
_tesseractOcrAutoFixes++;
labelFixesMade.Text = string.Format(" - {0}", _tesseractOcrAutoFixes);
LogOcrFix(listViewIndex, textWithOutFixes, line);
}
return line;
}
/// <summary>
/// Ocr via image compare
/// </summary>
/// <param name="bitmap">
/// The bitmap.
/// </param>
/// <param name="listViewIndex">
/// The list View Index.
/// </param>
/// <returns>
/// The <see cref="string"/>.
/// </returns>
private string SplitAndOcrBitmapNormalNew(Bitmap bitmap, int listViewIndex)
{
if (this._ocrFixEngine == null)
{
this.LoadOcrFixEngine(null, this.LanguageString);
}
string line = string.Empty;
var matches = new List<CompareMatch>();
var parentBitmap = new NikseBitmap(bitmap);
int minLineHeight = this._binOcrLastLowercaseHeight - 3;
if (minLineHeight < 5)
{
minLineHeight = this._nocrLastLowercaseHeight;
}
if (minLineHeight < 5)
{
minLineHeight = 6;
}
List<ImageSplitterItem> list = NikseBitmapImageSplitter.SplitBitmapToLettersNew(parentBitmap, (int)this.numericUpDownPixelsIsSpace.Value, this.checkBoxRightToLeft.Checked, Configuration.Settings.VobSubOcr.TopToBottom, minLineHeight);
int index = 0;
bool expandSelection = false;
bool shrinkSelection = false;
var expandSelectionList = new List<ImageSplitterItem>();
while (index < list.Count)
{
ImageSplitterItem item = list[index];
if (expandSelection || shrinkSelection)
{
expandSelection = false;
if (shrinkSelection && index > 0)
{
shrinkSelection = false;
}
else if (index + 1 < list.Count && list[index + 1].NikseBitmap != null)
{
// only allow expand to EndOfLine or space
index++;
expandSelectionList.Add(list[index]);
}
item = GetExpandedSelectionNew(parentBitmap, expandSelectionList);
this._vobSubOcrCharacter.Initialize(bitmap, item, this._manualOcrDialogPosition, this._italicCheckedLast, expandSelectionList.Count > 1, null, this._lastAdditions, this);
DialogResult result = this._vobSubOcrCharacter.ShowDialog(this);
this._manualOcrDialogPosition = this._vobSubOcrCharacter.FormPosition;
if (result == DialogResult.OK && this._vobSubOcrCharacter.ShrinkSelection)
{
shrinkSelection = true;
index--;
if (expandSelectionList.Count > 0)
{
expandSelectionList.RemoveAt(expandSelectionList.Count - 1);
}
}
else if (result == DialogResult.OK && this._vobSubOcrCharacter.ExpandSelection)
{
expandSelection = true;
}
else if (result == DialogResult.OK)
{
string text = this._vobSubOcrCharacter.ManualRecognizedCharacters;
string name = this.SaveCompareItemNew(item, text, this._vobSubOcrCharacter.IsItalic, expandSelectionList);
var addition = new ImageCompareAddition(name, text, item.NikseBitmap, this._vobSubOcrCharacter.IsItalic, listViewIndex);
this._lastAdditions.Add(addition);
matches.Add(new CompareMatch(text, this._vobSubOcrCharacter.IsItalic, expandSelectionList.Count, null));
expandSelectionList = new List<ImageSplitterItem>();
}
else if (result == DialogResult.Abort)
{
this._abort = true;
}
else
{
matches.Add(new CompareMatch("*", false, 0, null));
}
this._italicCheckedLast = this._vobSubOcrCharacter.IsItalic;
}
else if (item.NikseBitmap == null)
{
matches.Add(new CompareMatch(item.SpecialCharacter, false, 0, null));
}
else
{
CompareMatch bestGuess;
CompareMatch match = this.GetCompareMatchNew(item, out bestGuess, list, index);
if (match == null)
{
// Try line OCR if no image compare match
if (this._nOcrDb != null && this._nOcrDb.OcrCharacters.Count > 0)
{
match = this.GetNOcrCompareMatchNew(item, parentBitmap, this._nOcrDb, true, true);
}
}
if (match == null)
{
this._vobSubOcrCharacter.Initialize(bitmap, item, this._manualOcrDialogPosition, this._italicCheckedLast, false, bestGuess, this._lastAdditions, this);
DialogResult result = this._vobSubOcrCharacter.ShowDialog(this);
this._manualOcrDialogPosition = this._vobSubOcrCharacter.FormPosition;
if (result == DialogResult.OK && this._vobSubOcrCharacter.ExpandSelection)
{
expandSelectionList.Add(item);
expandSelection = true;
}
else if (result == DialogResult.OK)
{
string text = this._vobSubOcrCharacter.ManualRecognizedCharacters;
string name = this.SaveCompareItemNew(item, text, this._vobSubOcrCharacter.IsItalic, null);
var addition = new ImageCompareAddition(name, text, item.NikseBitmap, this._vobSubOcrCharacter.IsItalic, listViewIndex);
this._lastAdditions.Add(addition);
matches.Add(new CompareMatch(text, this._vobSubOcrCharacter.IsItalic, 0, null));
this.SetBinOcrLowercaseUppercase(item.NikseBitmap.Height, text);
}
else if (result == DialogResult.Abort)
{
this._abort = true;
}
else
{
matches.Add(new CompareMatch("*", false, 0, null));
}
this._italicCheckedLast = this._vobSubOcrCharacter.IsItalic;
}
else
{
// found image match
matches.Add(new CompareMatch(match.Text, match.Italic, 0, null));
if (match.ExpandCount > 0)
{
index += match.ExpandCount - 1;
}
}
}
if (this._abort)
{
return string.Empty;
}
if (!expandSelection && !shrinkSelection)
{
index++;
}
if (shrinkSelection && expandSelectionList.Count < 2)
{
shrinkSelection = false;
expandSelectionList = new List<ImageSplitterItem>();
}
}
line = GetStringWithItalicTags(matches);
if (this.checkBoxAutoFixCommonErrors.Checked && this._ocrFixEngine != null)
{
line = this._ocrFixEngine.FixOcrErrorsViaHardcodedRules(line, this._lastLine, null); // TODO: Add abbreviations list
}
if (this.checkBoxRightToLeft.Checked)
{
line = ReverseNumberStrings(line);
}
// OCR fix engine
string textWithOutFixes = line;
if (this._ocrFixEngine.IsDictionaryLoaded)
{
var autoGuessLevel = OcrFixEngine.AutoGuessLevel.None;
if (this.checkBoxGuessUnknownWords.Checked)
{
autoGuessLevel = OcrFixEngine.AutoGuessLevel.Aggressive;
}
if (this.checkBoxAutoFixCommonErrors.Checked)
{
line = this._ocrFixEngine.FixOcrErrors(line, listViewIndex, this._lastLine, true, autoGuessLevel);
}
int correctWords;
int wordsNotFound = this._ocrFixEngine.CountUnknownWordsViaDictionary(line, out correctWords);
if (wordsNotFound > 0 || correctWords == 0 || textWithOutFixes != null && string.IsNullOrWhiteSpace(textWithOutFixes.Replace("~", string.Empty)))
{
this._ocrFixEngine.AutoGuessesUsed.Clear();
this._ocrFixEngine.UnknownWordsFound.Clear();
line = this._ocrFixEngine.FixUnknownWordsViaGuessOrPrompt(out wordsNotFound, line, listViewIndex, bitmap, this.checkBoxAutoFixCommonErrors.Checked, this.checkBoxPromptForUnknownWords.Checked, true, autoGuessLevel);
}
if (this._ocrFixEngine.Abort)
{
this.ButtonStopClick(null, null);
this._ocrFixEngine.Abort = false;
return string.Empty;
}
// Log used word guesses (via word replace list)
foreach (string guess in this._ocrFixEngine.AutoGuessesUsed)
{
this.listBoxLogSuggestions.Items.Add(guess);
}
this._ocrFixEngine.AutoGuessesUsed.Clear();
// Log unkown words guess (found via spelling dictionaries)
this.LogUnknownWords();
if (wordsNotFound >= 3)
{
this.subtitleListView1.SetBackgroundColor(listViewIndex, Color.Red);
}
if (wordsNotFound == 2)
{
this.subtitleListView1.SetBackgroundColor(listViewIndex, Color.Orange);
}
else if (wordsNotFound == 1)
{
this.subtitleListView1.SetBackgroundColor(listViewIndex, Color.Yellow);
}
else if (string.IsNullOrWhiteSpace(line))
{
this.subtitleListView1.SetBackgroundColor(listViewIndex, Color.Orange);
}
else
{
this.subtitleListView1.SetBackgroundColor(listViewIndex, Color.LightGreen);
}
}
if (textWithOutFixes.Trim() != line.Trim())
{
this._tesseractOcrAutoFixes++;
this.labelFixesMade.Text = string.Format(" - {0}", this._tesseractOcrAutoFixes);
this.LogOcrFix(listViewIndex, textWithOutFixes, line);
}
return line;
}