提出了基于数据场的C均值聚类方法。引入数据场准确寻找初始中心的聚类点,根据势场分布函数,先找到k个势心,然后选择距离势心最近的样本数据,得到用数据场生成的初始中心,从而实现C均值聚类算法。该方法可以广泛地应用到空间数据的分类分级研究中。实例验证了该方法的可行性和科学性。 A C-means clustering method based on data field is proposed. According to the distribution function of potential field, k potential centers are found first, then the sample data closest to the potential center is selected to get the initial center generated by the data field, so as to achieve C-means clustering algorithm. The method can be widely applied to the classification and classification of spatial data. The example verifies the feasibility and scientificity of this method.