private void TrainLetter(ref int numberOfCharactersLeaned, ref int numberOfCharactersSkipped, BinaryOcrDb db, List <string> charactersLearned, string s, bool bold, bool italic)
{
Bitmap bmp = GenerateImageFromTextWithStyle(s, bold, italic);
var nbmp = new NikseBitmap(bmp);
nbmp.MakeTwoColor(280);
var list = NikseBitmapImageSplitter.SplitBitmapToLettersNew(nbmp, 10, false, false, 25);
if (list.Count == 1)
{
var match = db.FindExactMatch(new BinaryOcrBitmap(list[0].NikseBitmap));
if (match < 0)
{
pictureBox1.Image = list[0].NikseBitmap.GetBitmap();
labelInfo.Refresh();
Application.DoEvents();
db.Add(new BinaryOcrBitmap(list[0].NikseBitmap, false, 0, s, 0, 0));
charactersLearned.Add(s);
numberOfCharactersLeaned++;
labelInfo.Text = string.Format("Now training font '{1}', total characters learned is {0}, {2} skipped", numberOfCharactersLeaned, _subtitleFontName, numberOfCharactersSkipped);
bmp.Dispose();
}
else
{
numberOfCharactersSkipped++;
}
}
}
private void TrainLetter(ref int numberOfCharactersLeaned, ref int numberOfCharactersSkipped, NOcrDb nOcrD, List <string> charactersLearned, string s, bool bold)
{
Bitmap bmp = GenerateImageFromTextWithStyle(s, bold);
var nbmp = new NikseBitmap(bmp);
nbmp.MakeTwoColor(280);
var list = NikseBitmapImageSplitter.SplitBitmapToLettersNew(nbmp, 10, false, false, 25);
if (list.Count == 1)
{
NOcrChar match = nOcrD.GetMatch(list[0].NikseBitmap);
if (match == null)
{
pictureBox1.Image = list[0].NikseBitmap.GetBitmap();
this.Refresh();
Application.DoEvents();
System.Threading.Thread.Sleep(100);
NOcrChar nOcrChar = new NOcrChar(s);
nOcrChar.Width = list[0].NikseBitmap.Width;
nOcrChar.Height = list[0].NikseBitmap.Height;
VobSubOcrNOcrCharacter.GenerateLineSegments((int)numericUpDownSegmentsPerCharacter.Value, checkBoxVeryAccurate.Checked, nOcrChar, list[0].NikseBitmap);
nOcrD.Add(nOcrChar);
charactersLearned.Add(s);
numberOfCharactersLeaned++;
labelInfo.Text = string.Format("Now training font '{1}', total characters leaned is {0}, {2} skipped", numberOfCharactersLeaned, _subtitleFontName, numberOfCharactersSkipped);
bmp.Dispose();
}
else
{
numberOfCharactersSkipped++;
}
}
}
private void TrainLetter(ref int numberOfCharactersLearned, ref int numberOfCharactersSkipped, BinaryOcrDb db, string s, bool bold, bool italic)
{
var bmp = GenerateImageFromTextWithStyle("H " + s, bold, italic);
var nbmp = new NikseBitmap(bmp);
nbmp.MakeTwoColor(280);
var list = NikseBitmapImageSplitter.SplitBitmapToLettersNew(nbmp, 10, false, false, 25, false);
if (list.Count == 3)
{
var item = list[2];
var bob = new BinaryOcrBitmap(item.NikseBitmap, italic, 0, s, item.X, item.Y);
var match = db.FindExactMatch(bob);
if (match < 0)
{
db.Add(bob);
numberOfCharactersLearned++;
bmp.Dispose();
}
else
{
numberOfCharactersSkipped++;
}
}
}
internal void Initialize(Bitmap bitmap, int pixelsIsSpace, bool rightToLeft, NOcrDb nOcrDb, VobSubOcr vobSubOcr, double unItalicFactor)
{
_bitmap = bitmap;
var nbmp = new NikseBitmap(bitmap);
nbmp.ReplaceNonWhiteWithTransparent();
bitmap = nbmp.GetBitmap();
_bitmap2 = bitmap;
_nocrChars = nOcrDb.OcrCharacters;
_matchList = new List <VobSubOcr.CompareMatch>();
_vobSubOcr = vobSubOcr;
_unItalicFactor = unItalicFactor;
int minLineHeight = 6;
_imageList = NikseBitmapImageSplitter.SplitBitmapToLettersNew(nbmp, pixelsIsSpace, rightToLeft, Configuration.Settings.VobSubOcr.TopToBottom, minLineHeight);
// _imageList = NikseBitmapImageSplitter.SplitBitmapToLetters(nbmp, pixelsIsSpace, rightToLeft, Configuration.Settings.VobSubOcr.TopToBottom);
int index = 0;
while (index < _imageList.Count)
{
ImageSplitterItem item = _imageList[index];
if (item.NikseBitmap == null)
{
listBoxInspectItems.Items.Add(item.SpecialCharacter);
_matchList.Add(null);
}
else
{
nbmp = item.NikseBitmap;
nbmp.ReplaceNonWhiteWithTransparent();
item.Y += nbmp.CropTopTransparent(0);
nbmp.CropTransparentSidesAndBottom(0, true);
nbmp.ReplaceTransparentWith(Color.Black);
//get nocr matches
Nikse.SubtitleEdit.Forms.VobSubOcr.CompareMatch match = vobSubOcr.GetNOcrCompareMatchNew(item, nbmp, nOcrDb, _unItalicFactor, false, false);
if (match == null)
{
listBoxInspectItems.Items.Add("?");
_matchList.Add(null);
}
else
{
listBoxInspectItems.Items.Add(match.Text);
_matchList.Add(match);
}
}
index++;
}
}
internal void Initialize(Bitmap bitmap, int pixelsIsSpace, bool rightToLeft, NOcrDb nOcrDb, VobSubOcr vobSubOcr, bool italic, int minLineHeight, bool deepSeek)
{
_bitmap = bitmap;
var nikseBitmap = new NikseBitmap(bitmap);
nikseBitmap.ReplaceNonWhiteWithTransparent();
_nOcrChars = nOcrDb.OcrCharacters;
_nOcrDb = nOcrDb;
_matchList = new List <VobSubOcr.CompareMatch>();
_imageList = NikseBitmapImageSplitter.SplitBitmapToLettersNew(nikseBitmap, pixelsIsSpace, rightToLeft, Configuration.Settings.VobSubOcr.TopToBottom, minLineHeight, false);
int index = 0;
_indexLookup = new Dictionary <int, int>();
while (index < _imageList.Count)
{
var item = _imageList[index];
if (item.NikseBitmap == null)
{
_indexLookup.Add(listBoxInspectItems.Items.Count, index);
listBoxInspectItems.Items.Add(item.SpecialCharacter);
_matchList.Add(null);
}
else
{
var match = vobSubOcr.GetNOcrCompareMatchNew(item, nikseBitmap, nOcrDb, italic, deepSeek, index, _imageList);
if (match == null)
{
_indexLookup.Add(listBoxInspectItems.Items.Count, index);
listBoxInspectItems.Items.Add("?");
_matchList.Add(null);
}
else
{
_indexLookup.Add(listBoxInspectItems.Items.Count, index);
listBoxInspectItems.Items.Add(match.Text);
_matchList.Add(match);
if (match.ExpandCount > 0)
{
index += match.ExpandCount - 1;
}
}
}
index++;
}
}
private void TrainLetter(ref int numberOfCharactersLeaned, ref int numberOfCharactersSkipped, NOcrDb nOcrD, List <string> charactersLearned, string s, bool bold)
{
Bitmap bmp = GenerateImageFromTextWithStyle("H " + s, bold);
var nbmp = new NikseBitmap(bmp);
nbmp.MakeTwoColor(280);
nbmp.CropTop(0, Color.FromArgb(0, 0, 0, 0));
var list = NikseBitmapImageSplitter.SplitBitmapToLettersNew(nbmp, 10, false, false, 25);
if (list.Count == 3)
{
var item = list[2];
NOcrChar match = nOcrD.GetMatch(item.NikseBitmap, item.Top, false, false, 0);
if (match == null || match.Text != s)
{
labelInfo.Refresh();
Application.DoEvents();
NOcrChar nOcrChar = new NOcrChar(s)
{
Width = item.NikseBitmap.Width,
Height = item.NikseBitmap.Height,
MarginTop = item.Top,
};
VobSubOcrNOcrCharacter.GenerateLineSegments((int)numericUpDownSegmentsPerCharacter.Value, checkBoxVeryAccurate.Checked, nOcrChar, item.NikseBitmap);
nOcrD.Add(nOcrChar);
charactersLearned.Add(s);
numberOfCharactersLeaned++;
labelInfo.Text = string.Format("Now training font '{1}', total characters learned is {0:#,###,###}, {2:#,###,###} skipped", numberOfCharactersLeaned, _subtitleFontName, numberOfCharactersSkipped);
bmp.Dispose();
}
else
{
numberOfCharactersSkipped++;
}
}
}
private void BinaryOcrTrain_Shown(object sender, EventArgs e)
{
if (string.IsNullOrEmpty(_autoDetectFontText))
{
return;
}
BinaryOcrTrain_ResizeEnd(sender, e);
SelectAll_Click(this, EventArgs.Empty);
int numberOfCharactersLearned = 0;
int numberOfCharactersSkipped = 0;
foreach (ListViewItem fontItem in listViewFonts.CheckedItems)
{
_subtitleFontName = fontItem.Text;
labelInfo.Text = $"Checking font '{_subtitleFontName}'... {AutoDetectedFonts.Count} hits";
labelInfo.Refresh();
Application.DoEvents();
for (_subtitleFontSize = 20; _subtitleFontSize <= 100; _subtitleFontSize++)
{
if (!string.IsNullOrEmpty(_autoDetectFontText))
{
var s = _autoDetectFontText;
var bicDb = new BinaryOcrDb(null);
TrainLetter(ref numberOfCharactersLearned, ref numberOfCharactersSkipped, bicDb, s, false, false);
if (bicDb.FindExactMatch(_autoDetectFontBob) >= 0)
{
AutoDetectedFonts.Add(_subtitleFontName + " " + _subtitleFontSize);
}
else
{
// allow for error %
var smallestDifference = int.MaxValue;
foreach (var compareItem in bicDb.CompareImages)
{
if (compareItem.Width == _autoDetectFontBob.Width && compareItem.Height == _autoDetectFontBob.Height) // precise math in size
{
if (Math.Abs(compareItem.NumberOfColoredPixels - _autoDetectFontBob.NumberOfColoredPixels) < 3)
{
int dif = NikseBitmapImageSplitter.IsBitmapsAlike(compareItem, _autoDetectFontBob);
if (dif < smallestDifference)
{
if (!BinaryOcrDb.AllowEqual(compareItem, _autoDetectFontBob))
{
continue;
}
smallestDifference = dif;
if (dif < 3)
{
break; // foreach ending
}
}
}
}
}
if (smallestDifference < 13)
{
AutoDetectedFonts.Add($"{_subtitleFontName} {_subtitleFontSize} - diff={smallestDifference}");
}
}
bicDb = new BinaryOcrDb(null);
TrainLetter(ref numberOfCharactersLearned, ref numberOfCharactersSkipped, bicDb, s, false, true);
if (bicDb.FindExactMatch(_autoDetectFontBob) >= 0)
{
AutoDetectedFonts.Add(_subtitleFontName + " " + _subtitleFontSize + " italic");
}
if (checkBoxBold.Checked)
{
bicDb = new BinaryOcrDb(null);
TrainLetter(ref numberOfCharactersLearned, ref numberOfCharactersSkipped, bicDb, s, true, false);
if (bicDb.FindExactMatch(_autoDetectFontBob) >= 0)
{
AutoDetectedFonts.Add(_subtitleFontName + " " + _subtitleFontSize + " bold");
}
}
}
}
}
DialogResult = DialogResult.OK;
}
private void listBoxInspectItems_SelectedIndexChanged(object sender, EventArgs e)
{
labelImageInfo.Text = string.Empty;
labelExpandCount.Text = string.Empty;
if (listBoxInspectItems.SelectedIndex < 0)
{
return;
}
_selectedCompareNode = null;
_selectedCompareBinaryOcrBitmap = null;
var img = _imageSources[listBoxInspectItems.SelectedIndex];
pictureBoxInspectItem.Image = img;
if (img != null)
{
pictureBoxInspectItem.Width = img.Width + 2;
pictureBoxInspectItem.Height = img.Height + 2;
}
pictureBoxCompareBitmap.Image = null;
pictureBoxCompareBitmapDouble.Image = null;
int index = listBoxInspectItems.SelectedIndex;
var match = _matches[index];
_selectedMatch = match;
if (!string.IsNullOrEmpty(match.Name))
{
if (_binOcrDb != null)
{
bool bobFound = false;
foreach (BinaryOcrBitmap bob in _binOcrDb.CompareImages)
{
if (match.Name == bob.Key)
{
textBoxText.Text = bob.Text;
checkBoxItalic.Checked = bob.Italic;
_selectedCompareBinaryOcrBitmap = bob;
SetDbItemView(bob.ToOldBitmap());
var matchBob = new BinaryOcrBitmap(new NikseBitmap(_imageSources[listBoxInspectItems.SelectedIndex]));
if (matchBob.Hash == bob.Hash && matchBob.Width == bob.Width && matchBob.Height == bob.Height && matchBob.NumberOfColoredPixels == bob.NumberOfColoredPixels)
{
buttonAddBetterMatch.Enabled = false; // exact match
}
else
{
buttonAddBetterMatch.Enabled = true;
}
bobFound = true;
break;
}
}
if (!bobFound)
{
foreach (BinaryOcrBitmap bob in _binOcrDb.CompareImagesExpanded)
{
if (match.Name == bob.Key)
{
textBoxText.Text = bob.Text;
checkBoxItalic.Checked = bob.Italic;
_selectedCompareBinaryOcrBitmap = bob;
var oldBitmap = bob.ToOldBitmap();
SetDbItemView(oldBitmap);
int dif = 1;
if (oldBitmap.Width == match.ImageSplitterItem.NikseBitmap.Width && oldBitmap.Height == match.ImageSplitterItem.NikseBitmap.Height)
{
dif = NikseBitmapImageSplitter.IsBitmapsAlike(match.ImageSplitterItem.NikseBitmap, oldBitmap);
}
buttonAddBetterMatch.Enabled = dif > 0; // if exact match then don't allow "Add better match"
labelExpandCount.Text = $"Expand count: {bob.ExpandCount}";
break;
}
}
}
}
else
{
foreach (XmlNode node in ImageCompareDocument.DocumentElement.ChildNodes)
{
if (node.Attributes["Text"] != null && node.InnerText == match.Name)
{
string text = node.Attributes["Text"].InnerText;
textBoxText.Text = text;
checkBoxItalic.Checked = node.Attributes["Italic"] != null;
string databaseName = Path.Combine(_directoryPath, "Images.db");
using (var f = new FileStream(databaseName, FileMode.Open))
{
try
{
string name = node.InnerText;
int pos = Convert.ToInt32(name);
f.Position = pos;
var mbmp = new ManagedBitmap(f);
var bitmap = mbmp.ToOldBitmap();
SetDbItemView(mbmp.ToOldBitmap());
labelImageInfo.Text = string.Format(Configuration.Settings.Language.VobSubEditCharacters.Image + " - {0}x{1}", bitmap.Width, bitmap.Height);
}
catch (Exception exception)
{
labelImageInfo.Text = Configuration.Settings.Language.VobSubEditCharacters.Image;
MessageBox.Show(exception.Message);
}
}
_selectedCompareNode = node;
break;
}
}
}
}
buttonAddBetterMatch.Text = Configuration.Settings.Language.VobSubOcrCharacterInspect.AddBetterMatch;
if (_selectedMatch.Text == Configuration.Settings.Language.VobSubOcr.NoMatch)
{
buttonUpdate.Enabled = false;
buttonDelete.Enabled = false;
buttonAddBetterMatch.Enabled = true;
buttonAddBetterMatch.Text = Configuration.Settings.Language.VobSubOcrCharacterInspect.Add;
textBoxText.Enabled = true;
textBoxText.Text = string.Empty;
checkBoxItalic.Enabled = true;
pictureBoxCompareBitmap.Visible = true;
pictureBoxCompareBitmapDouble.Visible = true;
labelDoubleSize.Visible = true;
}
else if (_selectedCompareNode == null && _selectedCompareBinaryOcrBitmap == null)
{
buttonUpdate.Enabled = false;
buttonDelete.Enabled = false;
buttonAddBetterMatch.Enabled = true;
textBoxText.Enabled = true;
textBoxText.Text = string.Empty;
checkBoxItalic.Enabled = false;
pictureBoxCompareBitmap.Visible = false;
pictureBoxCompareBitmapDouble.Visible = false;
labelDoubleSize.Visible = false;
if (img == null)
{
buttonAddBetterMatch.Enabled = false;
}
}
else
{
buttonUpdate.Enabled = true;
buttonDelete.Enabled = true;
if (_selectedCompareNode != null)
{
buttonAddBetterMatch.Enabled = true;
}
textBoxText.Enabled = true;
checkBoxItalic.Enabled = true;
pictureBoxCompareBitmap.Visible = true;
pictureBoxCompareBitmapDouble.Visible = true;
labelDoubleSize.Visible = true;
}
}
private void TrainLetter(ref int numberOfCharactersLearned, ref int numberOfCharactersSkipped, NOcrDb nOcrD, string s, bool bold, bool italic, bool doubleLetter)
{
var bmp = GenerateImageFromTextWithStyle("H " + s, bold, italic);
var nikseBitmap = new NikseBitmap(bmp);
nikseBitmap.MakeTwoColor(280);
nikseBitmap.CropTop(0, Color.FromArgb(0, 0, 0, 0));
var list = NikseBitmapImageSplitter.SplitBitmapToLettersNew(nikseBitmap, 10, false, false, 25, false);
if (list.Count == 3)
{
var item = list[2];
var match = nOcrD.GetMatch(item.NikseBitmap, item.Top, false, 25);
if (match == null || match.Text != s)
{
labelInfo.Refresh();
Application.DoEvents();
var nOcrChar = new NOcrChar(s)
{
Width = item.NikseBitmap.Width,
Height = item.NikseBitmap.Height,
MarginTop = item.Top,
Italic = italic,
};
VobSubOcrNOcrCharacter.GenerateLineSegments((int)numericUpDownSegmentsPerCharacter.Value + (doubleLetter ? 20 : 0), false, nOcrChar, item.NikseBitmap);
nOcrD.Add(nOcrChar);
numberOfCharactersLearned++;
labelInfo.Text = string.Format(Configuration.Settings.Language.VobSubOcr.NowTraining, numberOfCharactersLearned, _subtitleFontName, numberOfCharactersSkipped);
bmp.Dispose();
}
else
{
numberOfCharactersSkipped++;
}
}
else if (!doubleLetter)
{
if (list.Count == 4 && list[2].NikseBitmap != null && list[3].NikseBitmap != null)
{
// e.g. quote (")
var expandItem = VobSubOcr.GetExpandedSelectionNew(nikseBitmap, new List <ImageSplitterItem> {
list[2], list[3]
});
var match = nOcrD.GetMatchExpanded(nikseBitmap, expandItem, 2, list);
if (match != null && match.Text == s)
{
numberOfCharactersSkipped++;
return;
}
var nOcrChar = new NOcrChar(s)
{
Width = expandItem.NikseBitmap.Width,
Height = expandItem.NikseBitmap.Height,
MarginTop = expandItem.Top,
Italic = italic,
ExpandCount = 2,
};
VobSubOcrNOcrCharacter.GenerateLineSegments((int)numericUpDownSegmentsPerCharacter.Value + 5, false, nOcrChar, expandItem.NikseBitmap);
nOcrD.Add(nOcrChar);
return;
}
if (list.Count == 5 && list[2].NikseBitmap != null && list[3].NikseBitmap != null && list[4].NikseBitmap != null)
{
// e.g. "%"
var expandItem = VobSubOcr.GetExpandedSelectionNew(nikseBitmap, new List <ImageSplitterItem> {
list[2], list[3], list[4]
});
var match = nOcrD.GetMatchExpanded(nikseBitmap, expandItem, 2, list);
if (match != null && match.Text == s)
{
numberOfCharactersSkipped++;
return;
}
var nOcrChar = new NOcrChar(s)
{
Width = expandItem.NikseBitmap.Width,
Height = expandItem.NikseBitmap.Height,
MarginTop = expandItem.Top,
Italic = italic,
ExpandCount = 3,
};
nOcrD.Add(nOcrChar);
VobSubOcrNOcrCharacter.GenerateLineSegments((int)numericUpDownSegmentsPerCharacter.Value + 10, false, nOcrChar, expandItem.NikseBitmap);
return;
}
numberOfCharactersSkipped++;
}
}
