Adaptive thresholding technique for document image analysis

Binarization of document images is mostly used as a preprocessing step in document image analysis. Proper binarization is very important so as to save all or maximum subcomponents such as text, background and picture. The work of Niblack is significant and has most acceptable results out of many techniques of thresholding. In this paper, we have presented a modified version of Niblack's method of thresholding (image binarization) that takes care of the limitations of Niblack's technique. Our algorithm works very well and is independent of many challenging factors involved in thresholding of document images like ambient illumination, contrast stretch, shading effects. An optimal size of window for local binarization scheme is also proposed based upon experiments and their results. The results show that the method adapts and perform well in each case.