Mean Shift and Fuzzy C-Means-Based Algorithm for Segmentation of Noisy Scenes

Abstract

In this paper, a segmentation algorithm has been proposed to deal with noisy brain MR images. The algorithm is a combination of mean shift algorithm and fuzzy c-means algorithm. Initially, the mean shift filter has been applied to eliminate noise followed by the application of fuzzy c-means algorithm for clustering. The mean shift algorithm with an appropriate kernel reinforced the mood seeking attribute of FCM algorithm to achieve proper clustering with noisy images. The algorithm has successfully been tested with general and brain MR images. The results have been compared with those of mean shift and fuzzy c-means algorithm.

Keywords: Clustering, Fuzzy c-means, Mean shift, Filtering