Fuzzy c-means Cluster Image Segmentation with Entropy Constraint

A Fuzzy c-means (FCM) cluster segmentation algorithm based on entropy constraint has been proposed to resolve problem of time wasting presented in traditional FCM algorithm. The minimum sample ratio under which the sampled image keeps most information of initial image was studied, and the limitation function was deduced. A relative entropy loss constraint based on histogram was introduced to keep sample image out of serious distortion and variable-step searching method was proposed to find out appropriate sample ratio. Experiments of single threshold was preformed and the results showed that the average time consuming of the proposed method is 2.9% of FCM method, 4.8% of 2D entropy method, and 6.6% of Otsu method, and the processing speed of the new method is increased by 10-120 times. The experiment results indicated that the new algorithm improves the processing efficiency of traditional FCM, and of cause can be applied to other kinds of FCM algorithm.