protected override RunInfo TesseractRunInfo(OcrParams ocrParams)
{
OcrMode mode = ocrParams.Mode;
string folder = mode == OcrMode.Fast || mode == OcrMode.Default ? "fast" : "best";
if (ocrParams.LanguageCode.Split('+').All(code => !File.Exists(Path.Combine(TesseractBasePath, folder, $"{code.ToLowerInvariant()}.traineddata"))))
{
// Use the other source if the selected one doesn't exist
folder = folder == "fast" ? "best" : "fast";
mode = folder == "fast" ? OcrMode.Fast : OcrMode.Best;
}
return(new RunInfo
{
Arguments = mode == OcrMode.Best ? "--oem 1" : mode == OcrMode.Legacy ? "--oem 0" : "",
DataPath = folder,
PrefixPath = folder
});
}
public static TessConvertResult BitmapToString(Bitmap bitmap, OcrMode mode)
{
string textResult;
float confidence;
if (bitmap.Size.IsEmpty)
{
return(null);
}
using (var engine = new TesseractEngine(@"./tessdata", "pol", EngineMode.Default))
{
if (mode == OcrMode.numeric)
{
engine.SetVariable("tessedit_char_whitelist", "-0123456789.,");
}
Tesseract.BitmapToPixConverter converter = new BitmapToPixConverter();
var imgPix = converter.Convert(bitmap);
imgPix = imgPix.ConvertRGBToGray();
imgPix = imgPix.Scale(7, 7);
//imgPix.Save("test.tiff");
using (var page = engine.Process(imgPix))
{
textResult = page.GetText();
confidence = page.GetMeanConfidence();
}
if (confidence == 0 || textResult == "")
{
return(null);
}
}
return(new TessConvertResult(textResult, confidence));
}
public OcrParams(string langCode, OcrMode mode)
{
LanguageCode = langCode;
Mode = mode;
}
