A Novel Unsupervised Segmentation Method for MR Brain Images Based on Fuzzy Methods

Image segmentation is an important research topic in image processing and computer vision community. In this paper, we present a novel segmentation method based on the combination of fuzzy connectedness and adaptive fuzzy C means (AFCM). AFCM handles intensity inhomogeneities problem in magnetic resonance images (MRI) and provides effective seeds for fuzzy connectedness simultaneously. With the seeds selected, fuzzy connectedness method is applied. As fuzzy connectedness method makes full use of the inaccuracy and ‘hanging togetherness’ property of MRI, our new method behaves well in both simulated and real images.