/// <summary>Preprocess given image if it is needed.</summary>
/// <param name="inputImage">
/// original input image
/// <see cref="System.IO.FileInfo"/>
/// </param>
/// <param name="pageNumber">number of page to be OCRed</param>
/// <returns>
/// path to output image as
/// <see cref="System.String"/>
/// </returns>
private String PreprocessImage(FileInfo inputImage, int pageNumber)
{
String tmpFileName = TesseractOcrUtil.GetTempFilePath(Guid.NewGuid().ToString(), GetExtension(inputImage));
String path = inputImage.FullName;
try {
if (GetTesseract4OcrEngineProperties().IsPreprocessingImages())
{
Pix pix = ImagePreprocessingUtil.PreprocessImage(inputImage, pageNumber);
TesseractOcrUtil.SavePixToTempPngFile(tmpFileName, pix);
if (!File.Exists(System.IO.Path.Combine(tmpFileName)))
{
System.Drawing.Bitmap img = TesseractOcrUtil.ConvertPixToImage(pix);
if (img != null)
{
TesseractOcrUtil.SaveImageToTempPngFile(tmpFileName, img);
}
}
}
if (!GetTesseract4OcrEngineProperties().IsPreprocessingImages() || !File.Exists(System.IO.Path.Combine(tmpFileName
)))
{
TesseractOcrUtil.CreateTempFileCopy(path, tmpFileName);
}
if (File.Exists(System.IO.Path.Combine(tmpFileName)))
{
path = tmpFileName;
}
}
catch (System.IO.IOException e) {
LogManager.GetLogger(GetType()).Error(MessageFormatUtil.Format(Tesseract4LogMessageConstant.CANNOT_READ_INPUT_IMAGE
, e.Message));
}
return(path);
}
public virtual void TestPreprocessingConditions()
{
Pix pix = null;
NUnit.Framework.Assert.IsNull(TesseractOcrUtil.ConvertToGrayscale(pix));
NUnit.Framework.Assert.IsNull(TesseractOcrUtil.OtsuImageThresholding(pix));
NUnit.Framework.Assert.IsNull(TesseractOcrUtil.ConvertPixToImage(pix));
TesseractOcrUtil.DestroyPix(pix);
}
/// <summary>
/// Reads input file as Leptonica
/// <see cref="Tesseract.Pix"/>
/// and
/// converts it to
/// <see cref="System.Drawing.Bitmap"/>.
/// </summary>
/// <param name="inputImage">
/// input image
/// <see cref="System.IO.FileInfo"/>
/// </param>
/// <returns>
/// returns a
/// <see cref="System.Drawing.Bitmap"/>
/// as the result
/// </returns>
internal static System.Drawing.Bitmap ReadAsPixAndConvertToBufferedImage(FileInfo inputImage)
{
Pix pix = Tesseract.Pix.LoadFromFile(inputImage.FullName);
return(TesseractOcrUtil.ConvertPixToImage(pix));
}
