A Fast Adaptive Binarization Method Based on Sub Block OSTU and Improved Sauvola

This paper has advanced a novel hybrid binarization method combining global thresholding and local adaptive thresholding used for foreground and background segmenting. In global thresholding, sub block OSTU is used to achieve better performance for it can represent local distribution characteristics while global OSTU method cannot. Original image and binary image are scanned individually to locate sub blocks which require further process. Sauvola algorithm implementing integral image is then performed for a second binarization. Simulation results on several printform envelopes show that this algorithm can maintain the same performance with the improved Sauvola algorithm while reducing the processing time by 27.4%.