Evaluation of Binarization Methods Using Global Threshold for Chinese Rubbing Images

Abstract

There are a huge amount of historical Chinese rubbing images in different library that have not been exploited electronically. However, many invaluable collections of rubbing images are already digitized and indexed for consulting, exchange and distant access purposes which protect them from direct manipulation. In many applications of digital rubbing image, the use of binary images can decrease the computational cost of the succeeding steps compared to using gray-level images. Thresholding is a simple but effective tool to separate characters from the background for rubbing image. In this paper, five global thresholding algorithms such as Histogram Isodata method, Minimum Cross Entropy(MCE) method, Kittler’s method, Maxium Entropy method,Ostu’s method and Fuzzy CMeans(FCM) clustering thresholding method that had been used widely by scholars have been researched. Two quantitative measure of comparison is provided by the Peak Signal to Noise Ratio (PSNR) and Mean Square Error (MSE) of the rubbing image for segmentation results. Keywords-Thresholding;Chinese rubbing image, Ostu’s thresholding; image segmentation; FCM