关键词: 图像分割, 进化聚类, 基于内容的遥感图像检索, 模糊C均值FCM聚类

Abstract: For there is no better way to make certain the initial each center of clusters and the value of C in fuzzy C-means method in the issue of image color clustering segmentation，a new method is introduced.Firstly，pixels of the image are segmented quickly based on the value of RGB by evolving clustering.At the same time the value of C is confirmed.Secondly the center of clustering that has been segmented is optimized by Fuzzy C-Means（FCM）.Finally，the image is segmented by FCM.The time complexity factor and space complexity factor of this method and the FCM are much the same.The FCM method is equivalent to this method which is proved by experiment.The method is applied to segment the remote sensing images and characters image.The results of the expriment indicate that this method is good at segmentation-accuracy and vague boundary region segmentation.

Key words: image segmentation, evolving clustering, Content-Based Remote Sensing Image Retrieval（CBRSIR）, Fuzzy C-Means clustering