Although several image thresholding algorithms based on minimum cross entropy criterion have been proposed in recent years

only the form of the criterion or a priori probability and conditional probability was employed. In this paper

a new algorithm based on maximum between-class cross entropy using a posterior probability is presented for image thresholding taken into account the dissimilarity between object and background in image. Suppose the conditional distributions of object and background are modeled with normal distributions

the a posterior probabilities are computed by Bayes formula. The new algorithm is compared with a number of traditional algorithms based on Shannon entropy and minimum cross entropy by applying them to various test images.