Abstract: In order to overcome the shortcoming of Fuzzy Cmeans (FCM) that cannot deal with image noise and weaknesses of its common improved algorithms, an improved FCM algorithm using the difference of neighborhood information was proposed in this paper, which used Gaussian function to characterize the difference of space and gray value about neighborhood pixels reasonably, and to adjust the center pixel membership to achieve the purpose of noise image segmentation. The experimental results show that the proposed algorithm can effectively deal with the images with Gaussian and Pepper & Salt noise, and can remove the noise while retaining more complete details of image. Its segmentation results are better than several improved FCM algorithms in the literatures.

Key words: Fuzzy C-Means (FCM), difference of space, difference of gray value, Gaussian function

摘要： 为了克服模糊C均值（FCM）无法处理图像噪声的缺点以及常用改进算法分割不足，提出了一种利用邻域差异性信息的FCM改进算法。利用高斯函数来合理刻画邻域间像素的空间位置和灰度差异特性，实现对中心像素隶属度的调整，达到分割噪声图像的目的。实验证明，该算法可以有效地处理高斯和椒盐噪声，在去除噪声的同时较完整地保留了图像的细节，其分割效果优于几种常用FCM改进算法。

关键词: 模糊C均值, 空间差异, 灰度值差异, 高斯函数