Modified FCM image segmentation based on Gaussian kernel and spatial information

摘要

This paper proposed a modified FCM image segmentation method based on Gaussian kernel and spatial information. FCM is especially effective in image segmentation, however, it lacks enough robustness to noise and outliers. At first, the Gaussian kernel and spatial information will be applied into the fuzzy clustering algorithm to enhance the robustness of FCM and reconstruct neighborhood information of the new objective function in this paper. Then the results of different methods are analyzed and compared. Experimental results show that this method can improve the image segmentation accuracy and robustness of anti-noise.