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 algorithm (FCM) is widely used in image processing and pattern recognition. This algorithm is essentially a local search optimization method, which is sensitive to the initial value and easy to fall into local minimum. When the number of clusters is relatively large, satisfactory clustering results are often not obtained. This paper first discusses the FCM algorithm initialization on the clustering results, and then proposes a morphological processing based FCM initialization method. This method can not only get more satisfactory results of clustering, but also automatically determine the number of clusters.