Abstract:

For the poor antinoise performance limitations of the traditional fuzzy C-means (FCM) algorithm, we proposed a new spatial fuzzy clustering optimization algorithm for image segmentation. We added a wealth of spatial information between pixels in the image feature items, so that the traditional FCM sensitive
to noise was solved, and the robustness of the algorithm was enhanced. Experimental results show that our algorithm can achieve the effectivesegmentation of the 5% Gaussian noise images, and the results are significantly better than those by traditional FCM image segmentation algorithm.

Key words: image segmentation, fuzzy clustering, FCM algorithm, spatial information