Mirrors/SubtitleEdit
1
0
Fork 0
mirror of https://github.com/SubtitleEdit/subtitleedit.git synced 2024-11-24 04:02:36 +01:00
Code Issues Projects Releases Wiki Activity

Work on OCR (minor optimize+fix type+work on train)

Browse Source
This commit is contained in:
Nikolaj Olsson 2020-05-22 21:05:37 +02:00
parent cdd8dd6363
commit a848d85b7d
3 changed files with 49 additions and 19 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

30
src/Forms/Ocr/BinaryOcrTrain.cs
View File

@ -25,6 +25,7 @@ namespace Nikse.SubtitleEdit.Forms.Ocr
private float _subtitleFontSize = 25.0f;
private readonly Color _borderColor = Color.Black;
private const float BorderWidth = 2.0f;
private bool _abort = false;
public BinaryOcrTrain()
{
@ -179,6 +180,12 @@ namespace Nikse.SubtitleEdit.Forms.Ocr
private void buttonTrain_Click(object sender, EventArgs e)
{
if (buttonTrain.Text == "Abort")
{
_abort = true;
return;
}
saveFileDialog1.DefaultExt = ".db";
saveFileDialog1.Filter = "*Binary Image Compare DB files|*.db";
if (saveFileDialog1.ShowDialog(this) != DialogResult.OK)
@ -186,6 +193,7 @@ namespace Nikse.SubtitleEdit.Forms.Ocr
return;
}
buttonTrain.Text = "Abort";
var startFontSize = Convert.ToInt32(comboBoxSubtitleFontSize.Items[comboBoxSubtitleFontSize.SelectedIndex].ToString());
var endFontSize = Convert.ToInt32(comboBoxFontSizeEnd.Items[comboBoxFontSizeEnd.SelectedIndex].ToString());
@ -235,11 +243,31 @@ namespace Nikse.SubtitleEdit.Forms.Ocr
labelInfo.Text = string.Format("Now training font '{1}', total characters learned is {0}", numberOfCharactersLeaned, _subtitleFontName);
}
}
if (_abort)
{
break;
}
}
if (_abort)
{
break;
}
}
if (_abort)
{
break;
}
}
bicDb.Save();
labelInfo.Text = "Training completed and saved in " + saveFileDialog1.FileName;
if (_abort)
{
labelInfo.Text = "Partially (aborted) training completed and saved in " + saveFileDialog1.FileName;
}
else
{
labelInfo.Text = "Training completed and saved in " + saveFileDialog1.FileName;
}
}
private void buttonOK_Click(object sender, EventArgs e)

36
src/Forms/Ocr/VobSubOcr.cs
View File

@ -3614,7 +3614,9 @@ namespace Nikse.SubtitleEdit.Forms.Ocr
return;
}
if (maxDiff < 0.2 || target.Width < 3 || target.Height < 5)
var tWidth = target.Width;
var tHeight = target.Height;
if (maxDiff < 0.2 || tWidth < 3 || tHeight < 5)
{
return;
}
@ -3624,7 +3626,7 @@ namespace Nikse.SubtitleEdit.Forms.Ocr
foreach (var compareItem in binOcrDb.CompareImages)
{
if (compareItem.Width == target.Width && compareItem.Height == target.Height) // precise math in size
if (compareItem.Width == tWidth && compareItem.Height == tHeight) // precise math in size
{
if (Math.Abs(compareItem.NumberOfColoredPixels - numberOfForegroundColors) < 3)
{
@ -3651,7 +3653,7 @@ namespace Nikse.SubtitleEdit.Forms.Ocr
{
foreach (var compareItem in binOcrDb.CompareImages)
{
if (compareItem.Width == target.Width && compareItem.Height == target.Height) // precise math in size
if (compareItem.Width == tWidth && compareItem.Height == tHeight) // precise math in size
{
if (Math.Abs(compareItem.NumberOfColoredPixels - numberOfForegroundColors) < 40)
{
@ -3675,11 +3677,11 @@ namespace Nikse.SubtitleEdit.Forms.Ocr
}
}
if (target.Width > 16 && target.Height > 16 && smallestDifference > 2) // for other than very narrow letter (like 'i' and 'l' and 'I'), try more sizes
if (tWidth > 16 && tHeight > 16 && smallestDifference > 2) // for other than very narrow letter (like 'i' and 'l' and 'I'), try more sizes
{
foreach (var compareItem in binOcrDb.CompareImages)
{
if (compareItem.Width == target.Width && compareItem.Height == target.Height - 1)
if (compareItem.Width == tWidth && compareItem.Height == tHeight - 1)
{
if (Math.Abs(compareItem.NumberOfColoredPixels - numberOfForegroundColors) < minForeColorMatch)
{
@ -3706,7 +3708,7 @@ namespace Nikse.SubtitleEdit.Forms.Ocr
{
foreach (var compareItem in binOcrDb.CompareImages)
{
if (compareItem.Width == target.Width && compareItem.Height == target.Height + 1)
if (compareItem.Width == tWidth && compareItem.Height == tHeight + 1)
{
if (Math.Abs(compareItem.NumberOfColoredPixels - numberOfForegroundColors) < minForeColorMatch)
{
@ -3734,7 +3736,7 @@ namespace Nikse.SubtitleEdit.Forms.Ocr
{
foreach (var compareItem in binOcrDb.CompareImages)
{
if (compareItem.Width == target.Width + 1 && compareItem.Height == target.Height + 1)
if (compareItem.Width == tWidth + 1 && compareItem.Height == tHeight + 1)
{
if (Math.Abs(compareItem.NumberOfColoredPixels - numberOfForegroundColors) < minForeColorMatch)
{
@ -3762,7 +3764,7 @@ namespace Nikse.SubtitleEdit.Forms.Ocr
{
foreach (var compareItem in binOcrDb.CompareImages)
{
if (compareItem.Width == target.Width - 1 && compareItem.Height == target.Height - 1)
if (compareItem.Width == tWidth - 1 && compareItem.Height == tHeight - 1)
{
if (Math.Abs(compareItem.NumberOfColoredPixels - numberOfForegroundColors) < minForeColorMatch)
{
@ -3790,7 +3792,7 @@ namespace Nikse.SubtitleEdit.Forms.Ocr
{
foreach (var compareItem in binOcrDb.CompareImages)
{
if (compareItem.Width - 1 == target.Width && compareItem.Height == target.Height)
if (compareItem.Width - 1 == tWidth && compareItem.Height == tHeight)
{
if (Math.Abs(compareItem.NumberOfColoredPixels - numberOfForegroundColors) < minForeColorMatch)
{
@ -3814,11 +3816,11 @@ namespace Nikse.SubtitleEdit.Forms.Ocr
}
}
if (smallestDifference > 9 && target.Width > 11)
if (smallestDifference > 9 && tWidth > 11)
{
foreach (var compareItem in binOcrDb.CompareImages)
{
if (compareItem.Width == target.Width - 2 && compareItem.Height == target.Height)
if (compareItem.Width == tWidth - 2 && compareItem.Height == tHeight)
{
if (Math.Abs(compareItem.NumberOfColoredPixels - numberOfForegroundColors) < minForeColorMatch)
{
@ -3842,11 +3844,11 @@ namespace Nikse.SubtitleEdit.Forms.Ocr
}
}
if (smallestDifference > 9 && target.Width > 14)
if (smallestDifference > 9 && tWidth > 14)
{
foreach (var compareItem in binOcrDb.CompareImages)
{
if (compareItem.Width == target.Width - 3 && compareItem.Height == target.Height)
if (compareItem.Width == tWidth - 3 && compareItem.Height == tHeight)
{
if (Math.Abs(compareItem.NumberOfColoredPixels - numberOfForegroundColors) < minForeColorMatch)
{
@ -3870,11 +3872,11 @@ namespace Nikse.SubtitleEdit.Forms.Ocr
}
}
if (smallestDifference > 9 && target.Width > 14)
if (smallestDifference > 9 && tWidth > 14)
{
foreach (var compareItem in binOcrDb.CompareImages)
{
if (compareItem.Width == target.Width && compareItem.Height == target.Height - 3)
if (compareItem.Width == tWidth && compareItem.Height == tHeight - 3)
{
if (Math.Abs(compareItem.NumberOfColoredPixels - numberOfForegroundColors) < minForeColorMatch)
{
@ -3898,11 +3900,11 @@ namespace Nikse.SubtitleEdit.Forms.Ocr
}
}
if (smallestDifference > 9 && target.Width > 14)
if (smallestDifference > 9 && tWidth > 14)
{
foreach (var compareItem in binOcrDb.CompareImages)
{
if (compareItem.Width - 2 == target.Width && compareItem.Height == target.Height)
if (compareItem.Width - 2 == tWidth && compareItem.Height == tHeight)
{
if (Math.Abs(compareItem.NumberOfColoredPixels - numberOfForegroundColors) < minForeColorMatch)
{

2
src/Logic/NikseBitmapImageSplitter.cs
View File

@ -771,7 +771,7 @@ namespace Nikse.SubtitleEdit.Logic
{
var list = new List<ImageSplitterItem>();
// split into seperate lines
// split into separate lines
List<ImageSplitterItem> verticalBitmaps = SplitVertical(bmp, minLineHeight, averageLineHeight);
if (!topToBottom)