ocrtesseractadaptive-threshold

What thresholding (binarization) algorithm is used in Tesseract OCR?

I am working on a project that needs accurate OCR results for images with rich background. So I am comparing results of two OCRs (one of them is Tesseract) to make my choice. The point is that results are strongly affected by the pre-processing step and especially image binarization. I extracted the binarized image of the other OCR and passed it to Tesseract which enhanced the results of Tesseract by 30-40%.

I have two questions and your answers would be of much help to me:

  1. What binarization algorithm does tesseract use, and is it configurable?
  2. Is there a way to extract the binarized image of Tesseract OCR so I can test the other OCR with it?

Thanks in advance :)

Solution

  • I think I have found the answers to my questions:

    1- The binarization algorithm used is Otsu thresholding. You can see it here in line 179.

    2- To get the binarized image, a method in tesseract api can be called:

    PIX* thresholded = api->GetThresholdedImage(); //thresholded must be freed