基于二维直方图的图像模糊聚类分割改进方法  

An Improved Fuzzy Clustering Algorithm for Image Segmentation Based on Two-Dimensional Histogram

作　　者：胡进生[1] 吴谨[1] 王兰[1] 

机构地区：[1]武汉科技大学信息科学与工程学院,湖北武汉430080

出　　处：《微计算机信息》2010年第5期210-211,129,共3页Control & Automation

基　　金：基金申请人:吴谨;项目名称:无源毫米波成像探测机理及关键技术研究;基金颁发部门:湖北省教育厅(D200711003)

摘　　要：基于二维直方图的模糊聚类分割方法,可以有效地抑制噪声的干扰。但是,FCM算法用于图像数据聚类的最大缺陷是运算开销太大,从而限制了该算法在图像分割中的应用。针对这个问题,本文将改进抑制式FCM算法拓展到二维空间,在保持空间信息的基础上,引入了自适应抑制因子来修正不同大小的隶属度,减少了数据的聚类次数,实验结果表明该方法提高了聚类的速度,又有很好的聚类效果。A fuzzy clustering algorithm for image segmentation based on two-dimensional histogram is robust for noise image. However, the fatal deficiency of FCM algorithm for image data clustering is high operational cost, which limits its application in image segmentation. To solve this problem, in the paper the improved suppressed FCM is extended to the two-dimensional space and the degree of membership is amended by adaptive inhibitory factor, which induce the number of data clustering while maintaining spatial information. Experimental results indicate that the new alzorithm increases the speed of clustering and has a good clustering effect.

关 键 词：图像分割 FCM 改进抑制式FCM 二维直方图 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]