Abstract: Fuzzy C means clustering is a classic non-supervised clustering model, successfully applied to remote sensing classification. However, the method is the sensitivity to the initial values selected randomly, easy to fall into a local optimal solution; also uses only spectral information and ignores spatial information. This paper presents a new clustering algorithm integrates with Fuzzy C-means clustering and Markov random field. The density function of the first principal component sufficiently reflects the class differences, from which the initial label for FCM algorithm can be efficiently determined, and the sensitivity of the initial value selected at random can be avoided. Meanwhile, this algorithm takes into account the spatial location information between pixels. The experiment shows that the proposed method is better than the general FCM algorithm.