关键词: 模糊C均值聚类, 二维直方图, 熵, 遥感图像, 图像分割

Abstract: A new effective multi-thresholds image segmentation method based on two-dimensional histogram FCM & entropy clustering is presented.Fuzzy C-means clustering algorithm has been widely used in automated image segmentation.However，the conventional FCM algorithm is noise sensitive because of not taking account of the spatial information.Fuzzy C-means clustering algorithm based on two-dimensional histogram is robust for noise，because it utilizes the gray level information of each pixel and its spatial correlation information within the neighborhood.The entropy term is introduced in object function that can suppress noise effectively and reduce influence of estimation of the cluster centers.Experimental results indicate that the new algorithm is better than FCM algorithm.

Key words: fuzzy C-means clustering, two-dimensional histogram, entropy, remote sensing image, image segmentation