A multiobjective fuzzy clustering algorithm based on robust local spatial information for image segmentation

Abstract

To obtain the satisfying performance of noisy image segmentation, a multiobjective fuzzy clustering algorithm based on robust local spatial information (MFC-RLS) is proposed. In this method, the robust local spatial information derived from the image is introduced into fitness functions which utilize the fuzzy compactness and fuzzy separation among the clusters. In addition, after producing the set of non-dominated solutions, the final segmentation result is chosen by a validity index with the robust local spatial information. Experimental results show that MFC-RLS behaves well in segmenting noisy images. © 2013 Springer-Verlag Berlin Heidelberg.