Abstract: Fuzzy exponent entropy based on exponent entropy is defined and its properties is proved，avoiding the problems of logarithmic behavior.The fuzzy patition definition of an image is given by the conditional probability and the concept，automatic image segmentaion algorithm using the maximum entropy is presented.For reducing computation complexity and enhancing speed，the paper designs self-adaptive similartaxis operator and random dissimilation operator with small probability.The improved Mind Evolutionary Algorithm（MEA） optimizes the fuzzy parameters and searches the optimal threshold value for image segmentation.The experimental results show that the proposed method can select the threshold values automatically and efficiently，better than Otsu and other two methods.