模糊c-均值算法改进及其对卫星遥感数据聚类的对比  

Improved Fuzzy C-mean Classifier and Comparison Study of Its Clustering Results of Satellite Remotely Sensed Data

作　　者：哈斯巴干[1] 马建文[1] 李启青[1] 刘志丽[1] 韩秀珍[1] 

机构地区：[1]中国科学院遥感应用研究所,北京100101

出　　处：《计算机工程》2004年第11期14-15,91,共3页Computer Engineering

基　　金：遥感数据智能处理技术与集成项目(CX020014);奥运科技专项项目(2002BA904B07)

摘　　要：提出的改进的模糊c-均值聚类方法采用基于标准协方差矩阵的Mahalanobis距离,即椭球体聚类方法,这种聚类算法更接近遥感数据散点图的实际情况,从而可以显著提高聚类效果。对北京卫星ASTER数据的聚类分析实验表明,改进的模糊c-均值聚类方法的聚类效果要优于K-均值聚类方法和常规的模糊c-均值聚类方法。This article presents an improved fuzzy c-mean classifier in which Mahalanobis distance is adopted using standard connivance matrix, shown as ellipse spheroid cluster algorithm. This methodis more likely close to remote sensing data scatter map then that of other cluster algorithm so that the classification results are better either. Satellite ASTER Beijing data is used for testing the results proved that the improved the Mahalanobis distance classifier is more precedence than k-mean classifier and fuzzy c-mean classifier.

关 键 词：遥感数据 K-均值聚类 模糊C均值聚类 MAHALANOBIS距离 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]