MR image segmentation algorithm based on non-local fuzzy C-means clustering

Fuzzy C-means clustering (FCM) is a kind of popular image segmentation algorithm currently. In this paper, Robust Fuzzy C-means clustering (RFCM) was improved by adding non-local weight value in it as penalty term, using redundant information of image to eliminate the influence of noise. Therefore, an image segmentation algorithm based on non-local fuzzy C-means clustering has been proposed. Clinical MR brain images were segmented to verify the algorithm, the experimental results shows that: for segmentation of brain tissue in noisy MR images, the OLR overlap rate may reach above 90%. The algorithm has good performance for noisy MR brain image segmentation.