private void DoCommandShowOptions()
{
DlgOptions dlg = new DlgOptions();
dlg.DataPath = _tessData;
if (dlg.ShowDialog(this) == DialogResult.OK)
{
_tessData = dlg.DataPath;
_lang = dlg.Language;
_ocrEngineMode = (int)dlg.OcrEngineMode;
bool status = _ocrProcessor.Init(_tessData, _lang, _ocrEngineMode);
Console.WriteLine(string.Format("[DEBUG] Init status: {0}", status));
string msg = string.Format("{0} to initialize Tesseract Engine {1}.",
(status ? "Succeed" : "Failed"), _ocrProcessor.GetTesseractEngineVersion());
MessageBox.Show(msg);
}
}
public string UseTesseract(string imgFile)
{
this.ErrMsg = string.Empty;
string defLang = "eng";
string strResult = string.Empty;
try
{
using (TesseractProcessor processor = new TesseractProcessor())
{
processor.Init(this.TESSDATA, defLang, this.oem);
processor.SetPageSegMode((ePageSegMode)Enum.Parse(typeof(ePageSegMode), this.pageSegMode));
#if DEBUG
System.Diagnostics.Debug.WriteLine("processor:");
System.Diagnostics.Debug.WriteLine(processor.GetTesseractEngineVersion());
#endif
string strIndicate = Path.GetFileNameWithoutExtension(imgFile);
strResult = processor.Recognize(imgFile);
if (!string.IsNullOrEmpty(strResult))
{
// correct common errors caused by OCR
strResult = this.CorrectOCRErrors(strResult);
// correct letter cases
strResult = this.CorrectLetterCases(strResult);
}
}
strResult = strResult.Replace("\n", Environment.NewLine);
}
catch (Exception ex)
{
this.ErrMsg = ex.ToString();
strResult = string.Empty;
}
return strResult;
}