聚类分析是多元统计分析的一种，也是无监督模式识别的一个重要分支，在模式分类 图像处理和模糊规则处理等众多领域中获得最广泛的应用。它把一个没有类别标记的样本按照某种准则划分为若干子集，使相似的样本尽可能归于一类，而把不相似的样本划分到不同的类中。硬聚类把每个待识别的对象严格的划分某类中，具有非此即彼的性质，而模糊聚类建立了样本对类别的不确定描述，更能客观的反应客观世界，从而成为聚类分析的主流。-Clustering analysis is a kind of multivariate statistical analysis, unsupervised pattern recognition is an important branch in pattern classification, image processing and fuzzy rule processing and so on themultitudinous domain gain the most widely used. It puts a no classes according to certain standards of sample mark divided into several subsets, make similar samples as far as possible, and not to class of similar sample divided into different classes. Hard clustering each to identify the object strict division, which has certain either/or properties, and fuzzy clustering established sample of categories of uncertainty descr iption, more can objectively reaction to the objective world, becoming the mainstream of the cluster analysis.