A comparison of binarization methods for historical archive documents

This paper compares several alternative binarization algorithms for historical archive documents, by evaluating their effect on end-to-end word recognition performance in a complete archive document recognition system utilising a commercial OCR engine. The algorithms evaluated are: global thresholding; Niblack´s and Sauvola´s algorithms; adaptive versions of Niblack´s and Sauvola´s algorithms; and Niblack´s and Sauvola´s algorithms applied to background removed images. We found that, for our archive documents, Niblack´s algorithm can achieve better performance than Sauvola´s (which has been claimed as an evolution of Niblack´s algorithm), and that it also achieved better performance than the internal binarization provided as part of the commercial OCR engine.