中文摘要：

针对模糊c-均值聚类方法对初始值敏感,且在聚类时忽略空间相关信息的不足提出一种基于马尔可夫随机场的模糊c-均值聚类方法,该方法用马尔可夫随机场来描述像元的空间相关性,形成顾及空间相关的模糊c-均值分类方法。初始值依据第一主成分的密度函数确定,既克服对初始值的依赖性,又在聚类的时候考虑空间相关信息。通过实例数据验证,所提出的方法分类精度优于传统的模糊c-均值模型。

英文摘要：

Fuzzy c-means（FCM） clustering is a classic unsupervised clustering model,which has been successfully applied to remote sensing classification.However,the method is sensitive to the initial values selected randomly,thus easy to reach a local optima;also it considers only spectral information and ignores spatial information.A clustering algorithm is proposed which integrates FCM clustering with Markov random field.The density function based on the first principal component which sufficiently reflects the class differences is used to determine the initial labels for FCM algorithm,thus the sensitivity to the random initial value can be avoided.Meanwhile,this algorithm takes into account the spatial information between pixels.The experiment shows that the new method is better than the general FCM algorithm.