Abstract：The classical Fuzzy C-Means clustering algorithm (FCM) is widely used in image segmentation．However FCM clustering algorithm usually leads to local minimum result and noise sensitive．Skull stripping on human brain by using seeds filling and morphological operations is proposed．The original Euclidean distance in the FCM is replaced by a kernel-induced distance，by using kernel method．Then，spatial information is added to the objective function．The experimental results show that the new algorithm can segment the image effectively and properly，and it is shown to be more robust to noises．
2010, Vol. 1 