public IEnumerable<RecognizedSegment> Recognize(Stream document, IProgressReporter progress)
{
var imageData = new ImageData(document);
var boards = imageData.DefineIteratedBoards();
var heuristics = boards.Segment().ToList();
progress.Maximum = heuristics.Count * 1000;
foreach (var segment in heuristics) {
var whitespaceResults = trainingData.PerformWhitespaceLookup(segment, progress.ScaledChildOperation(500)).LastOrDefault(r => r.Certainty > 10);
if (whitespaceResults != null && whitespaceResults.Text == "AllLabels") {
var match = trainingData.PerformLookup(segment, progress.ScaledChildOperation(500)).LastOrDefault(r => r.Certainty > 10);
if (match != null)
yield return match;
} else
progress.Progress += 500; //Add the progress that would have been used by the character recognition
}
}
public void TestAlgorithm()
{
string renderText = "أدخل نص هنا لترجمة";
//string renderText = "أد";
//string renderText = "ﺀﺁﺂﺃﺄﺅﺆﺇﺈﺉﺊﺋﺌﺍﺎﺏﺐﺑﺒﺓﺔﺕﺖﺗﺘﺙﺚﺛﺜﺝﺞﺟﺠﺡﺢﺣﺤﺥﺦﺧﺨﺩﺪﺫﺬﺭﺮﺯﺰﺱﺲﺳﺴﺵﺶﺷﺸﺹﺺﺻﺼﺽﺾﺿﻀﻁﻂﻃﻄﻅﻆﻇﻈﻉﻊﻋﻌﻍﻎﻏﻐﻑﻒﻓﻔﻕﻖﻗﻘﻙﻚﻛﻜﻝﻞﻟﻠﻡﻢﻣﻤﻥﻦﻧﻨﻩﻪﻫﻬﻭﻮﻯﻰﻱﻲﻳﻴﻵﻶﻷﻸﻹﻺﻻﻼ";
//string renderText = "ﺬﻤﺈﺗﻹﻕﻐﻘﻝﺺﻔﺒﺭﻧﻂﺓﻫﻲﺸﺪﺊﻭﻜﺶﻆﻸﻒﺆﺄﻤﻴﺣﻄﻗﻖﺹﺶﺅﺆﺔﻣﻕﻨﺝﻓﻂﺭﺡﺀﻷﺣﺜﻷﻞﺶﺼﺈﻎﻍﺁﺂﻊﻓﻶﻃﺰﻊﻒﺑﺈﻙﻦﻬﺣﻃﺲﺄﺿﻠﺄﻊﺰﺆﺊﻁﻗﺨﻣﺡﺚﺱﻤﺴﺄﺑﺙﺂﻮﻖﺢﻴﻘﺐﻴﺞﺹﺬﻔﺛﺄﺷﻅﻫﻈﺳﻎﻃﻉﺵﻨﺇﻤﺱﻴﻔﻱﺲﻨﺑﺩﺯﺒﻥﺽﺍﻯﺠﺏﻨﺇﺫﻑﻈﺜﻟﻨﻶﻨﻎﻀﻒﺵﺹﻫﺚﺏﺧﺙﺟﺪﺇﻁﻂﻮﻑﺤﻝﺭﺻﻁﺮﺴﻝﺭﺽﺳﺛﺔﺔﻩﻸﻆﺙﻎﻓﻶﻖﻷﺣﻺﻇﺇﻬﺾﺀﻦﺎﻖﻈﺻﺋﻈﺭﺌﺑﺞﻕﺋﺮﻤﻱﺒﺅﺳﺮﺽﺨﻱﺛﻗﻊﻣﺊﺽﻶﺧﻄﺞﺭﻔﺤﻁﻉﻕﻝﺯﺘﺌﺼﺴﺡﻊﻈﺼﺉﺵﺁﺹﺏﺿﺾﺚﺻﻭﻭﺥﺽﻬﺓﻧﻗﺷﻚﻗﺿﺯﻅﺬﻒﻼﺥﺛﺴﻣﺶﻼﺚﻋﺳﻁﺥﺊﺎﻫﺕﺊﺆﻙﻥﺸﻯﺨﻶﻒﻚﻧﻭﻮﻹﻗ";
var output = new DrawingGroup();
var format = new BasicTextParagraphProperties("Times New Roman", 14, FlowDirection.LeftToRight);
var words = BoundedWord.GetWords(renderText, Measurer.MeasureLines(renderText, 200, format, output)).ToList();
//using (var stream = output.ToBitmap().CreateStream()) {
using (var stream = File.OpenRead(@"C:\Users\SSL\Temp\Arabic.png")) {
//using (var targetStream = File.Create(@"C:\Users\SSL\Temp\Arabic-Console.png"))
// stream.CopyTo(targetStream);
stream.Position = 0;
var imageData = new ImageData(stream);
stream.Position = 0;
//System.Drawing.Bitmap b = boards.Boards.First().Matrix.ExtractRectangularContentArea(segment.Bounds).ConvertDoubleArrayToBitmap(System.Drawing.Color.White);
//whitespaceResults.Last().Log(boards.Boards.First().Matrix.ExtractRectangularContentArea(segment.Bounds).ConvertDoubleArrayToBitmap(System.Drawing.Color.White));
//characterResults.Last().Log(boards.Boards.First().Matrix.ExtractRectangularContentArea(segment.Bounds).ConvertDoubleArrayToBitmap(System.Drawing.Color.White));
//Debug.Print(characterResults.Last().Text + " " + characterResults.Last().Certainty.ToString());
var trainingData = new MutableReferenceSet();
trainingData.ReadFrom(trainingFolder);
var searcher = new ReferenceSearcher(trainingData);
var recognizer = new OriginalDocumentRecognizer(searcher);
var reporter = new ConsoleProgressReporter(false);
var results = recognizer.Recognize(stream, reporter).ToList().AsReadOnly();
var outputString = new StreamReader(OutputRenderer.PlainText.Convert(stream, results)).ReadToEnd();
Debug.Print(outputString);
}
}
//TODO: Test for white space first
public void TrainAlgorithm()
{
string renderText = "أدخل نص هنا لترجمة";
//string renderText = "ﻂﺎﺉﻳﺕﺍﻈﺕﻊﺹﻫﻑﻈﻂﻟﺐﺻﺔﺛﺅﺡﺽﻛﻟﺊﺒﻖﺼﻞﺫﻳﺀﺎﻠﻜﻯﺛﺠﻔﺡﺜﺠﻹﺡﺕﺨﺜﻛﻃﺜﺗﺯﺘﺈﻞﺄﺰﺦﻨﺂﺳﺋﺳﺟﻃﻪﻃﻤﻣﻭﻱﺚﺽﻷﻦﺁﻁﺹﻊﻞﻲﻷﻶﻁﺓﻃﻥﺅﺂﻗﺿﻲﺵﻑﻪﺐﻩﻚﻛﺧﻢﺵﻵﻓﺨﻹﻯﺝﺟﻝﺿﻳﻘﺉﺌﺸﺒﻮﺋﻼﺅﺽﻢﺋﺭﻻﻢﻨﺍﻕﺔﻤﻛﺢﻹﺽﺂﺣﺙﺖﻑﻇﻣﺢﺘﻈﺙﺫﻄﺱﻺﻷﻎﻛﺢﺥﻏﻩﺥﺩﺗﺇﺐﻆﻈﻘﺘﺶﺍﻌﺶﻬﻕﻱﺫﺺﻖﻀﻰﻢﻡﺶﺣﻘﻗﺙﺖﻼﻼﻔﻫﺕﻺﻬﻼﻉﺶﺜﻖﺼﺰﻠﺞﻵﻸﺎﺢﻙﺯﻤﻋﺏﻵﺿﻄﻝﻡﺵﻬﺯﻂﺘﺶﺵﻞﺠﻮﻥﻱﻟﺄﺺﻦﺳﺲﺁﺘﺪﻜﻉﻴﺹﻍﺘﺉﻛﺜﺻﺧﺟﻥﺁﺭﺧﻷﺗﺪﻥﺗﻗﺜﺹﻥﻍﻠﺐﺥﻂﺍﻃﻅﺶﻕﺣﺸﺈﻇﺞﺬﺂﺃﺳﺽﻒﻶﺃﺌﺿﺸﺚﻓﺯﺒﺄﺬﺺﻐﺀﺴﻵﺶﻗﺭﻡﻗﺮﺨﺊﻫﻯﻌﺌﺹﺕﻔﻦﻸﺒﻓﻝﺄﻢﺉﻯﻬﺕﺤﻐﻳﻺﺵﻇﻦﺴﻉﺨﺅﻙﻷﻇﻗﺂﻐﻆﻊﻇﻲﻺﺭﻃﺾﺽﺩﻓﻰﻐﻎﺺﻱﻁﺊﻕﺓﻭﺺﺅﺢﺄﻈﺫﻠﺟﺘﺌﺆﺂﺑﺿﺤﺤﺟﺦﺣﺰﺼﻴﺞﺒﻀﻶﻗﻇﻤﺶﺼﺂﻫﻟﺌﻅﺥﻮﺱﺁﻟﺟﺥﺝﻍﻮﺾﺎﺖﻌﺡﻠﺱﻠﺱﻂﻣﺏﻁﻏﻐﺘﺷﻞﻌﺃﺓﺮﺳﻜﻼﺮﻲﺡﻍﺝﻎﺀﻂﻎﺷﻉﺱﺘﺔﻙﻍﺊﺮﻆﻞﺕﺭﻞﺭﻇﻤﻃﻍﻉﻛﺏﻗﺶﻍﻯﻫﺐﺒﺔﻱﻢﻊﻯﺗﺇﻣﺞﻉﻰﺳﺉﺀﻩﻺﻦﺁﺓﻰﺆﺵﺊﺰﻵﻰﺨﻮﻷﻈﺢﻪﺽﻴﺚﻒﻸﻈﺺﻆﺦﻠﻅﻶﻚﺖﻑﺏﻫﺹﻣﻮﻗﺜﺌﻛﻧﺲﻒﻻﻬﺕﻨﻓﺧﺅﻳﺑﻣﺗﺁﺞﻄﻉﻉﺢﻬﺐﻒﻞﺞﺘﺃﺾﻡﺱﺒﺠﻨﺻﻨﻅﻲﺻﻗﺑﻵﻯﻏﻔﺟﻀﺩﻓﺷﻄﺩﺎﻐﻦﻬﻖﻝﻰﻜﻓﺒﺪﻮﻳﻰﻓﺦﺧﺬﺸﺨﺈﻯﺸﺪﻺﺾﻺﺾﻛﻬﺬﺆﺝﺇﺄﻍﺁﻋﺑﻥﺂﻜﺃﻻﻊﺀﻘﻡﻉﺚﺁﺧﻻﺛﺡﺛﺗﺼﻯﺘﻠﺱﺁﻤﻣﺣﻑﻬﻫﺆﺤﻘﺚﻐﺦﺩﻬﺺﻜﻺﻢﺘﻲﻀﻫﺄﺇﺁﻢﻦﺵﻁﻙﻶﺄﻍﻚﺛﺘﺧﺫﻨﻳﻃﻲﺥﺌﺫﻱﻵﺫﻚﻜﻲﺲﻇﺼﺽﻵﻞﺛﻐﺊﺅﻍﻠﻻﺡﻌﻖﻋﻐﺡﺵﻎﻷﺢﺮﻋﻅﻴﻙﺌﻩﺺﻘﻢﻤﻃﺹﺐﺜﺣﺷﺋﻘﻞﻭﻳﺒﺦﻨﻜﺛﺪﻘﺸﺧﺣﺸﺥﺎﻯﺰﻦﺿﻃﺝﻈﻱﻺﺗﻚﺒﻅﺯﺭﻧﺩﻳﺘﺲﺷﺸﺞﺌﻺﻭﺿﻞﺠﺺﻁﺋﺔﺂﺩﻯﻍﻉﺕﻀﺤﺂﺔﺏﺍﻊﻓﻷﺪﻍﻻﺭﺟﻰﻓﺷﺒﻚﻂﻒﻛﻦﻑﻀﻓﻏﻺﻂﺅﺺﺚﻝﻞﻕﺍﺵﺣﺎﻼﺰﺚﻄﺴﻏﺼﺓﻨﺡﺫﻯﻝﻰﺇﺿﻢﺻﺮﻑﺹﺻﻌﺐﻲﻬﺳﺝﻌﻇﻢﺞﺓﺃﻛﺒﺸﺗﺲﺁﺃﺿﻣﻘﺖﺒﻚﺦﻖﺞﻯﻏﺇﻏﻒﻁﻠﺉﻎﻄﺛﺞﻄﺡﺼﺍﻜﻚﺕﻲﺇﻊﺈﺬﻍﻷﻔﻱﻷﻠﺒﺌﻄﺭﻔﺓﺖﻀﺮﺙﺮﺯﻅﻫﻢﻢﻷﻂﺜﺉﺳﺀﺑﻡﺗﺥﻆﻘﺁﻗﻩﻫﺏﻫﺲﺎﻗﺗﻉﺨﻳﺛﺫﻸﻃﻕﺅﺚﻆﻨﻼﺑﻔﻭﺬﺑﻸﻠﺖﺯﻉﺻﺾﻓﺭﻀﺟﻖﺸﻔﺊﻠﻶﻰﺒﺗﻑﺖﺝﺿﺟﺬﺑﺖﻗﻘﺖﺞﺾ";
var output = new DrawingGroup();
var format = new BasicTextParagraphProperties("Times New Roman", 14, FlowDirection.LeftToRight);
var words = BoundedWord.GetWords(renderText, Measurer.MeasureLines(renderText, 200, format, output)).ToList();
var stream = output.ToBitmap().CreateStream();
var imageData = new ImageData(stream);
imageData.SaveFile("RenderedFile1.png");
stream.Close();
Directory.CreateDirectory(trainingFolder);
var boards = imageData.DefineIteratedBoards();
var trainingData = new MutableReferenceSet();
//ADD TO EXISTING TRAINING DATA LIBRARY:
//trainingData.ReadFrom(trainingFolder);
var characters = words.SelectMany(w => w.Characters).ToList();
for (int i = 0; i < characters.Count(); i++) {
var ch1 = characters[i];
var letterHeuristics = boards.GetLetterHeuristics(ch1);
if (i < characters.Count() - 1) {
var ch2 = characters[i + 1];
var whitespaceHeuristics = boards.GetSpaceHeuristics(ch1, ch2);
trainingData.AddHeuristics(whitespaceHeuristics);
}
if (letterHeuristics != null)
trainingData.AddHeuristics(letterHeuristics);
}
ImageUtilities.Utilities.TrainingLog.ToString();
trainingData.WriteTo(trainingFolder);
}