opencv

opencv / ImageBinarization / 0.1.1

README.md

This algorithm binarizes the input image, useful for cleaning background noise in a text image to be input to OCR. This algorithm uses the implementation described in http://liris.cnrs.fr/christian.wolf/software/binarize/

The parameters are identical to those explained in the website. Unless the user finds a compelling reason, in general, the newest algorithm, "w", is the best performing algorithm and it is the default. You may wish to experiment with different parameters to find out the one that best fits your image.

         -n Niblack (1986)       needs white text on black background
         -s Sauvola et al. (1997) needs black text onwhite background
         -w Wolf et al. (2001)   needs black text on white background

Default version: w (Wolf et al. 2001)
Default value for "k": 0.5

example:
      ["data://opencv/SampleImages/sampleOCR.jpg", 
       "", "data://opencv/SampleImages/sampleOCRCleaned.jpg"]