public override string RecognizeText(IList <Image> images, string lang)
{
string tessdata = Path.Combine(basedir, TESSDATA);
using (TesseractEngine engine = new TesseractEngine(tessdata, lang, EngineMode.Default))
{
engine.SetVariable("tessedit_create_hocr", Hocr ? "1" : "0");
Tesseract.PageSegMode psm = (PageSegMode)Enum.Parse(typeof(PageSegMode), PageSegMode);
StringBuilder strB = new StringBuilder();
int pageNum = 0;
foreach (Image image in images)
{
pageNum++;
using (Pix pix = ConvertBitmapToPix(image))
{
using (Page page = engine.Process(pix, psm))
{
string text = Hocr ? page.GetHOCRText(pageNum - 1) : page.GetText();
if (text == null)
{
return(String.Empty);
}
strB.Append(text);
}
}
}
return(strB.ToString().Replace("\n", Environment.NewLine));
}
}
protected override void LoadRegistryInfo(RegistryKey regkey)
{
base.LoadRegistryInfo(regkey);
selectedPSM = (string)regkey.GetValue(strPSM, Enum.GetName(typeof(PageSegMode), Tesseract.PageSegMode.Auto));
try
{
// validate PSM value
Tesseract.PageSegMode psm = (PageSegMode)Enum.Parse(typeof(PageSegMode), selectedPSM);
}
catch
{
selectedPSM = Enum.GetName(typeof(PageSegMode), Tesseract.PageSegMode.Auto);
}
}
/// <summary>
/// 识别白底黑字的二值化字符图片
/// </summary>
/// <param name="bmp">白底黑字的二值化字符图片</param>
/// <param name="mode">页面分析模式</param>
/// <returns>识别出的字符串</returns>
protected virtual string CrackCodeCore(Bitmap bmp, PageSegMode mode)
{
if (_engine == null)
{
throw new InvalidOperationException("Tesseract识别引擎 未初始化");
}
Tesseract.PageSegMode tmode = mode.CastTo(Tesseract.PageSegMode.Auto);
using (Page page = _engine.Process(bmp, tmode))
{
string text = page.GetText().Replace("\n", "").Replace(" ", "");
if (mode == PageSegMode.SingleBlockVertText)
{
text = text.ReverseString();
}
return(text);
}
}
/// <summary>
/// Processes a file using ResultRenderers.
/// </summary>
/// <param name="renderer"></param>
/// <param name="filename"></param>
private void ProcessImageFile(IResultRenderer renderer, string filename)
{
IEnumerable <string> configs_file = new List <string>()
{
CONFIGS_FILE
};
using (TesseractEngine engine = new TesseractEngine(Datapath, Language, EngineMode, configs_file))
{
ControlParameters(engine);
Tesseract.PageSegMode psm = (PageSegMode)Enum.Parse(typeof(PageSegMode), PageSegMode);
var imageName = Path.GetFileNameWithoutExtension(filename);
using (var pixA = LoadPixArray(filename))
{
using (renderer.BeginDocument(imageName))
{
foreach (var pix in pixA)
{
Pix pixd = null;
try
{
if (Deskew)
{
pixd = pix.Deskew(new ScewSweep(range: 45), Pix.DefaultBinarySearchReduction, Pix.DefaultBinaryThreshold, out Scew scew);
}
using (var page = engine.Process(pixd ?? pix, imageName, psm))
{
var addedPage = renderer.AddPage(page);
}
}
finally
{
if (pixd != null)
{
((IDisposable)pixd).Dispose();
}
}
}
}
}
}
}
public override string RecognizeText(IList <Image> images, string inputName)
{
IEnumerable <string> configs_file = new List <string>()
{
CONFIGS_FILE
};
using (TesseractEngine engine = new TesseractEngine(Datapath, Language, EngineMode, configs_file))
{
engine.SetVariable("tessedit_create_hocr", OutputFormat == "hocr" ? "1" : "0");
ControlParameters(engine);
Tesseract.PageSegMode psm = (PageSegMode)Enum.Parse(typeof(PageSegMode), PageSegMode);
StringBuilder strB = new StringBuilder();
int pageNum = 0;
foreach (Image image in images)
{
pageNum++;
using (Pix pix = ConvertBitmapToPix(image))
{
using (Page page = engine.Process(pix, inputName, psm))
{
string text = OutputFormat == "hocr" ? page.GetHOCRText(pageNum - 1) : page.GetText();
if (text == null)
{
return(String.Empty);
}
strB.Append(text);
}
}
}
return(strB.ToString().Replace("\n", Environment.NewLine));
}
}
