基于空间加权距离的自适应Fuzzy C-Means算法研究

An Adaptive Fuzzy C-Means Clustering Algorithm Based on Spatial Weighted Distance

空间聚类不仅应考虑GIS对象属性特征的相似性,还应考虑对象的空间邻近性。不同属性、位置特征在聚类中起到的作用不同。采用信息熵方法计算空间距离中各属性距离、位置距离的权重,权值大小用于度量相应特征在fuzzy c-means隶属度计算时的作用大小,并引入相似性指标,当两个聚类之间的相似度高于某个合并阈值时,则对应的一对聚类进行合并,从而克服需预先设置聚类类数的问题。通过应用实例的聚类有效性分析,与普通空间距离相比,基于空间加权距离的FCM算法具有稳定性和有效性。

Spatial clustering should not only consider the similarity of attributes features of GIS objects , but also consider spatial prox-imity of objects .Different attributes and location features play different roles in the clustering .Entropy method is used to calculate the weight of each attribute , location distance , which can measure the effect size of corresponding feature when fuzzy c -means member-ship is calculated.Moreover, fuzzy similarity index is used to assess the similarity of two clusters and similar clusters will be merged if the similarity between clusters is higher than a threshold , which can avoid presetting the number of clusters .The new algorithm is il-lustrated and analyzed by cluster validity indices and the result indicates it is more robust and effective for GIS data than the original FCM algorithm based on normal spatial distance .