Abstract
For improving the accuracy of the traditional maximum entropy threshold segmentation algorithm, an improved maximum entropy segmentation algorithm is proposed. Firstly, it determines the possible range of an optimal segmentation threshold according to a simple statistical method, so as to reduce the interference of the background and magnify the proportion of the target region. Secondly, in a certain range of threshold, does image segmentation according to an optimal segmentation threshold, which is obtained by using maximum entropy principle. Simulation experiments show that the improved algorithm not only can improve accuracy and noise immunity effectively, but also can better keep the details of the target region in comparison with the traditional maximum entropy threshold segmentation algorithm.