Fuzzy c-means clustering algorithm with deformable spatial information for image segmentation

Due to the fuzzy c-means(FCM) clustering algorithm is very sensitive to noise and outliers, the spatial information derived from neighborhood window is often used to improve its image segmentation performance. However, the geometric structures of neighborhood window are usually fixed for each pixel. This may affect the quality of spatial information. In this paper, a deformable strategy is presented to address this problem. The proposed strategy defines a novel neighborhood window with free deformation form, whose shape can be adjusted adaptively for each pixel. Specifically, the offset is introduced for each pixel within the neighborhood window to obtain the deformable neighborhood window. The offset can be learned in each iteration of FCM. By using the proposed deformable strategy, the FCM algorithms with deformable spatial information can be easily developed based on previous FCM algorithms with spatial information. Those deformable spatial information based FCM algorithms perform well than their original variants on noisy images. In the meantime, the ability of image details preservation of fuzzy local information c-means clustering algorithm (FLICM) is significantly improved by using the deformable strategy. The experiment results of six spatial information based FCM algorithms show that the proposed deformable strategy is very effective.