Mirrors/SubtitleEdit
1
0
Fork 0
mirror of https://github.com/SubtitleEdit/subtitleedit.git synced 2024-10-28 06:52:35 +01:00
Code Issues Projects Releases Wiki Activity

Working on ocr

Browse Source
This commit is contained in:
nikse.dk 2014-02-20 21:35:26 +01:00
parent 0228ffce04
commit 5252c9f970
4 changed files with 153 additions and 66 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

31
src/Forms/VobSubEditCharacters.cs
View File

@ -16,7 +16,6 @@ namespace Nikse.SubtitleEdit.Forms
List<bool> _italics = new List<bool>();
internal List<VobSubOcr.ImageCompareAddition> Additions { get; private set; }
BinaryOcrDb _binOcrDb = null;
List<BinaryOcrBitmap> _binOcrListLookups = null;
public XmlDocument ImageCompareDocument
{
@ -103,6 +102,18 @@ namespace Nikse.SubtitleEdit.Forms
}
}
}
foreach (BinaryOcrBitmap bob in _binOcrDb.CompareImagesExpanded)
{
string name = bob.Key;
foreach (VobSubOcr.ImageCompareAddition a in additions)
{
if (name == a.Name && bob.Text != null)
{
listBoxFileNames.Items.Add(bob);
_italics.Add(bob.Italic);
}
}
}
}
else
{
@ -154,6 +165,13 @@ namespace Nikse.SubtitleEdit.Forms
texts.Add(text);
count++;
}
foreach (BinaryOcrBitmap bob in _binOcrDb.CompareImagesExpanded)
{
string text = bob.Text;
if (!texts.Contains(text) && text != null)
texts.Add(text);
count++;
}
}
else
{
@ -199,6 +217,15 @@ namespace Nikse.SubtitleEdit.Forms
_italics.Add(bob.Italic);
}
}
foreach (BinaryOcrBitmap bob in _binOcrDb.CompareImagesExpanded)
{
string text = bob.Text;
if (text == target)
{
listBoxFileNames.Items.Add(bob);
_italics.Add(bob.Italic);
}
}
}
else
{
@ -257,7 +284,9 @@ namespace Nikse.SubtitleEdit.Forms
{
var bob = GetSelectedBinOcrBitmap();
if (bob != null)
{
bmp = bob.ToOldBitmap();
}
}
else if (File.Exists(databaseName))
{

109
src/Forms/VobSubOcr.cs
View File

@ -2501,61 +2501,36 @@ namespace Nikse.SubtitleEdit.Forms
return null;
}
var bob = new BinaryOcrBitmap(target);
// build merged items - for expanded match search
var mergedItems = new List<ImageSplitterItem>();
List<ImageSplitterItem> expandSelectionList = new List<ImageSplitterItem>();
for (int j = listIndex; j < list.Count; j++)
for (int k = 0; k < _binaryOcrDb.CompareImagesExpanded.Count; k++)
{
ImageSplitterItem item = list[j];
if (item.NikseBitmap == null)
var b = _binaryOcrDb.CompareImagesExpanded[k];
if (bob.Hash == b.Hash && bob.Width == b.Width && bob.Height == b.Height && bob.NumberOfColoredPixels == b.NumberOfColoredPixels)
{
break;
}
else
{
expandSelectionList.Add(item);
if (expandSelectionList.Count > 1 && expandSelectionList.Count < 10)
mergedItems.Add(GetExpandedSelectionNew(parentBitmap, expandSelectionList));
}
}
// check for expand match!
foreach (BinaryOcrBitmap compareItem in _binaryOcrDb.CompareImages)
{
if (compareItem.ExpandCount > 0 &&
mergedItems.Count > 0 &&
compareItem.Width > target.Width &&
parentBitmap.Width >= compareItem.Width + targetItem.X &&
parentBitmap.Height >= compareItem.Height)
{
foreach (ImageSplitterItem merged in mergedItems)
bool ok = false;
for (int i = 0; i < b.ExpandedList.Count; i++)
{
if (merged.NikseBitmap.Width == compareItem.Width && merged.NikseBitmap.Height == compareItem.Height)
if (listIndex + i + 1 < list.Count && list[listIndex + i + 1].NikseBitmap != null && b.ExpandedList[i].Hash == new BinaryOcrBitmap(list[listIndex + i + 1].NikseBitmap).Hash)
{
int dif = NikseBitmapImageSplitter.IsBitmapsAlike(compareItem, merged.NikseBitmap);
if (dif < smallestDifference)
{
bool allow = true;
if (Math.Abs(target.Height - compareItem.Height) > 5 && compareItem.Text == "\"")
allow = false;
if (allow)
{
smallestDifference = dif;
smallestIndex = index;
if (dif == 0)
break; // foreach ending
}
}
ok = true;
}
else
{
ok = false;
break;
}
}
if (ok)
{
secondBestGuess = null;
return new CompareMatch(b.Text, b.Italic, b.ExpandCount, b.Key);
}
}
index++;
}
// Search images with minor location changes
FindBestMatchNew(ref index, ref smallestDifference, ref smallestIndex, target, _binaryOcrDb);
FindBestMatchNew(ref index, ref smallestDifference, ref smallestIndex, target, _binaryOcrDb, bob);
if (maxDiff > 0)
{
if (smallestDifference * 100.0 / (target.Width * target.Height) > _vobSubOcrSettings.AllowDifferenceInPercent && target.Width < 70)
@ -2563,20 +2538,20 @@ namespace Nikse.SubtitleEdit.Forms
if (smallestDifference > 2 && target.Width > 25)
{
var cutBitmap = target.CopyRectangle(new Rectangle(4, 0, target.Width - 4, target.Height));
FindBestMatchNew(ref index, ref smallestDifference, ref smallestIndex, cutBitmap, _binaryOcrDb);
FindBestMatchNew(ref index, ref smallestDifference, ref smallestIndex, cutBitmap, _binaryOcrDb, bob);
double differencePercentage = smallestDifference * 100.0 / (target.Width * target.Height);
}
if (smallestDifference > 2 && target.Width > 12)
{
var cutBitmap = target.CopyRectangle(new Rectangle(1, 0, target.Width - 2, target.Height));
FindBestMatchNew(ref index, ref smallestDifference, ref smallestIndex, cutBitmap, _binaryOcrDb);
FindBestMatchNew(ref index, ref smallestDifference, ref smallestIndex, cutBitmap, _binaryOcrDb, bob);
}
if (smallestDifference > 2 && target.Width > 12)
{
var cutBitmap = target.CopyRectangle(new Rectangle(0, 0, target.Width - 2, target.Height));
FindBestMatchNew(ref index, ref smallestDifference, ref smallestIndex, cutBitmap, _binaryOcrDb);
FindBestMatchNew(ref index, ref smallestDifference, ref smallestIndex, cutBitmap, _binaryOcrDb, bob);
}
if (smallestDifference > 2 && target.Width > 12)
@ -2585,7 +2560,7 @@ namespace Nikse.SubtitleEdit.Forms
int topCrop = 0;
var cutBitmap2 = NikseBitmapImageSplitter.CropTopAndBottom(cutBitmap, out topCrop, 2);
if (cutBitmap2.Height != target.Height)
FindBestMatchNew(ref index, ref smallestDifference, ref smallestIndex, cutBitmap2, _binaryOcrDb);
FindBestMatchNew(ref index, ref smallestDifference, ref smallestIndex, cutBitmap2, _binaryOcrDb, bob);
}
if (smallestDifference > 2 && target.Width > 15)
@ -2594,7 +2569,7 @@ namespace Nikse.SubtitleEdit.Forms
int topCrop = 0;
var cutBitmap2 = NikseBitmapImageSplitter.CropTopAndBottom(cutBitmap, out topCrop);
if (cutBitmap2.Height != target.Height)
FindBestMatchNew(ref index, ref smallestDifference, ref smallestIndex, cutBitmap2, _binaryOcrDb);
FindBestMatchNew(ref index, ref smallestDifference, ref smallestIndex, cutBitmap2, _binaryOcrDb, bob);
}
if (smallestDifference > 2 && target.Width > 15)
@ -2603,7 +2578,7 @@ namespace Nikse.SubtitleEdit.Forms
int topCrop = 0;
var cutBitmap2 = NikseBitmapImageSplitter.CropTopAndBottom(cutBitmap, out topCrop);
if (cutBitmap2.Height != target.Height)
FindBestMatchNew(ref index, ref smallestDifference, ref smallestIndex, cutBitmap2, _binaryOcrDb);
FindBestMatchNew(ref index, ref smallestDifference, ref smallestIndex, cutBitmap2, _binaryOcrDb, bob);
}
}
}
@ -2655,9 +2630,8 @@ namespace Nikse.SubtitleEdit.Forms
return bmp;
}
private static void FindBestMatchNew(ref int index, ref int smallestDifference, ref int smallestIndex, NikseBitmap target, BinaryOcrDb binOcrDb)
private static void FindBestMatchNew(ref int index, ref int smallestDifference, ref int smallestIndex, NikseBitmap target, BinaryOcrDb binOcrDb, BinaryOcrBitmap bob)
{
var bob = new BinaryOcrBitmap(target);
var bobExactMatch = binOcrDb.FindExactMatch(bob);
if (bobExactMatch >= 0)
{
@ -3369,12 +3343,29 @@ namespace Nikse.SubtitleEdit.Forms
return name;
}
private string SaveCompareItemNew(NikseBitmap newTarget, string text, bool isItalic, int expandCount)
private string SaveCompareItemNew(NikseBitmap newTarget, string text, bool isItalic, List<ImageSplitterItem> expandList)
{
BinaryOcrBitmap bob = new BinaryOcrBitmap(newTarget, isItalic, expandCount, text);
_binaryOcrDb.Add(bob);
_binaryOcrDb.Save();
return bob.Key;
int expandCount = 0;
if (expandList != null)
expandCount = expandList.Count;
if (expandCount > 0)
{
var bob = new BinaryOcrBitmap(expandList[0].NikseBitmap, isItalic, expandCount, text);
bob.ExpandedList = new List<BinaryOcrBitmap>();
for (int j = 1; j < expandList.Count; j++)
bob.ExpandedList.Add(new BinaryOcrBitmap(expandList[j].NikseBitmap));
_binaryOcrDb.Add(bob);
_binaryOcrDb.Save();
return bob.Key;
}
else
{
var bob = new BinaryOcrBitmap(newTarget, isItalic, expandCount, text);
_binaryOcrDb.Add(bob);
_binaryOcrDb.Save();
return bob.Key;
}
}
/// <summary>
@ -3632,7 +3623,7 @@ namespace Nikse.SubtitleEdit.Forms
else if (result == DialogResult.OK)
{
string text = _vobSubOcrCharacter.ManualRecognizedCharacters;
string name = SaveCompareItemNew(item.NikseBitmap, text, _vobSubOcrCharacter.IsItalic, expandSelectionList.Count);
string name = SaveCompareItemNew(item.NikseBitmap, text, _vobSubOcrCharacter.IsItalic, expandSelectionList);
var addition = new ImageCompareAddition(name, text, item.NikseBitmap, _vobSubOcrCharacter.IsItalic, listViewIndex);
_lastAdditions.Add(addition);
matches.Add(new CompareMatch(text, _vobSubOcrCharacter.IsItalic, expandSelectionList.Count, null));
@ -3670,7 +3661,7 @@ namespace Nikse.SubtitleEdit.Forms
else if (result == DialogResult.OK)
{
string text = _vobSubOcrCharacter.ManualRecognizedCharacters;
string name = SaveCompareItemNew(item.NikseBitmap, text, _vobSubOcrCharacter.IsItalic, 0);
string name = SaveCompareItemNew(item.NikseBitmap, text, _vobSubOcrCharacter.IsItalic, null);
var addition = new ImageCompareAddition(name, text, item.NikseBitmap, _vobSubOcrCharacter.IsItalic, listViewIndex);
_lastAdditions.Add(addition);
matches.Add(new CompareMatch(text, _vobSubOcrCharacter.IsItalic, 0, null));

2
src/Logic/OCR/Binary/BinaryOcrBitmap.cs
View File

@ -1,4 +1,5 @@
using System;
using System.Collections.Generic;
using System.Drawing;
using System.IO;
@ -27,6 +28,7 @@ namespace Nikse.SubtitleEdit.Logic.OCR.Binary
public int ExpandCount { get; set; }
public bool LoadedOK { get; private set; }
public string Text { get; set; }
public List<BinaryOcrBitmap> ExpandedList { get; set; }
public string Key
{

77
src/Logic/OCR/Binary/BinaryOcrDb.cs
View File

@ -8,6 +8,7 @@ namespace Nikse.SubtitleEdit.Logic.OCR.Binary
{
public string FileName { get; private set; }
public List<BinaryOcrBitmap> CompareImages = new List<BinaryOcrBitmap>();
public List<BinaryOcrBitmap> CompareImagesExpanded = new List<BinaryOcrBitmap>();
public BinaryOcrDb(string fileName)
{
@ -29,6 +30,12 @@ namespace Nikse.SubtitleEdit.Logic.OCR.Binary
{
foreach (var bob in CompareImages)
bob.Save(gz);
foreach (var bob in CompareImagesExpanded)
{
bob.Save(gz);
foreach (var ExpandedBob in bob.ExpandedList)
ExpandedBob.Save(gz);
}
gz.Flush();
gz.Close();
}
@ -38,6 +45,7 @@ namespace Nikse.SubtitleEdit.Logic.OCR.Binary
public void LoadCompareImages()
{
var list = new List<BinaryOcrBitmap>();
var expandList = new List<BinaryOcrBitmap>();
if (!File.Exists(FileName))
{
@ -54,18 +62,39 @@ namespace Nikse.SubtitleEdit.Logic.OCR.Binary
{
var bob = new BinaryOcrBitmap(gz);
if (bob.LoadedOK)
list.Add(bob);
{
if (bob.ExpandCount > 0)
{
expandList.Add(bob);
bob.ExpandedList = new List<BinaryOcrBitmap>();
for (int i = 1; i < bob.ExpandCount; i++)
{
var expandedBob = new BinaryOcrBitmap(gz);
if (expandedBob.LoadedOK)
bob.ExpandedList.Add(expandedBob);
else
break;
}
}
else
{
list.Add(bob);
}
}
else
{
done = true;
}
}
}
}
CompareImages = list;
CompareImagesExpanded = expandList;
}
public int FindExactMatch(BinaryOcrBitmap bob)
{
for (int i=0; i<CompareImages.Count; i++)
for (int i = 0; i < CompareImages.Count; i++)
{
var b = CompareImages[i];
if (bob.Hash == b.Hash && bob.Width == b.Width && bob.Height == b.Height && bob.NumberOfColoredPixels == b.NumberOfColoredPixels)
@ -74,13 +103,49 @@ namespace Nikse.SubtitleEdit.Logic.OCR.Binary
return -1;
}
public int FindExactMatchExpanded(BinaryOcrBitmap bob)
{
for (int i = 0; i < CompareImagesExpanded.Count; i++)
{
var b = CompareImagesExpanded[i];
if (bob.Hash == b.Hash && bob.Width == b.Width && bob.Height == b.Height && bob.NumberOfColoredPixels == b.NumberOfColoredPixels)
return i;
}
return -1;
}
public int Add(BinaryOcrBitmap bob)
{
int index = FindExactMatch(bob);
if (index == -1)
CompareImages.Add(bob);
int index;
if (bob.ExpandCount > 0)
{
index = FindExactMatchExpanded(bob);
if (index == -1 || CompareImagesExpanded[index].ExpandCount != bob.ExpandCount)
{
CompareImagesExpanded.Add(bob);
}
else
{
bool allAlike = true;
for (int i=0; i < bob.ExpandCount-1; i++)
{
if (bob.ExpandedList[i].Hash != CompareImagesExpanded[index].ExpandedList[i].Hash)
allAlike = false;
}
if (!allAlike)
CompareImages.Add(bob);
else
System.Windows.Forms.MessageBox.Show("Expanded image already in db!");
}
}
else
System.Windows.Forms.MessageBox.Show("Image already in db!");
{
index = FindExactMatch(bob);
if (index == -1)
CompareImages.Add(bob);
else
System.Windows.Forms.MessageBox.Show("Image already in db!");
}
return index;
}