摘要
模糊C均值聚类算法(FCM)在图像处理和模式识别中有着广泛的应用,该算法实质上是一种局部搜索寻优方法,对初始值很敏感,容易陷入局部极小值。当聚类数比较多时,往往得不到满意的聚类结果。本文首先讨论了FCM算法初始化对聚类结果的影响,然后提出了一种基于形态处理的FCM初始化方法。这种方法不仅可以得到比较满意的聚类结果,而且可以自动确定聚类数。
Fuzzy c─means clustering(FCM ) algorithm is widely applied in variety areas such as image processing and pattern recognition The algorithm is essentially a partially searching optimization. It is very sensitive to the initial、value and it often gets partial minimum results.When there are more clusters,satisfied rcsults can’t be often getted. Therefore,the paper first discusses the influence of the initialization of FCM,then presents a initialization method based on Binarv MorPholcgy. Not only can the approach obtain satisfied results but also can get the number of clus ters automaically.
出处
《系统工程与电子技术》
EI
CSCD
1995年第11期64-69,共6页
Systems Engineering and Electronics
基金
国家自然科学基金
关键词
FCM
模糊聚类分析
模式识别
算法
Fuzzy c─means clustering alorithm, lnitialization,Morphology