This paper introduces a new image segmentation method in the framework of Shadowed C-Means clustering. By implanting the local spatial information in the estimation procedure of membership values and mapping the original data into a high dimensional Hilbert space, we propose the Kernel Spatial Shadowed C-Means (KSSCM) clustering algorithm. Compared with traditional Fuzzy C-Means and Shadowed C-Means based approaches, the KSSCM based image segmentation can obtain better results on synthetic and real test images. We observed that the proposed KSSCM can effectively tackle the overlapping among segments and suppress the noise in images. KSSCM is an efficient approach for noise image segmentation.