结合空间信息的FCM聚类噪声图像分割方法

摘要

The standard fuzzy C-means clustering algorithm dose not consider any spatial information, so it is highly sensitive to noise and no good to noise image segmentation. In order to overcome this shortcoming, a fuzzy C-means clustering algorithm using spatial information for image segmentation is proposed. The new algorithm incorporates spatial information into the similarity measure and the membership function. This paper considers that spatial information depends on the relative location and features of the neighbour pixels. Experimental results show that this method is effective in some noisy image segmentation, and has better performance of resisting noise.%标准模糊C均值聚类算法由于没有考虑任何与图像空间连续性有关的信息,对噪声高度敏感,针对这一问题,提出一种基于图像空间信息的FCM聚类分割算法.该算法将图像像素的空间信息引入到相似性度量和隶属度函数中,其中空间信息由像素的相对位置和邻域内像素的特征决定.实验结果证明,该方法能有效地对含有一定噪声的图像进行分割,具有较好的抗噪性能.