Chan-Vese Model Image Segmentation with Neighborhood Information

Intensity inhomogeneous images,especially medical images,widely exist in real world.Considering that the existing methods can not solve intensity inhomogeneous images segmentation properly,this paper proposes an improved Chan-Vese model with neighborhood information.Firstly,the pixels of the same kind with the central pixel were obtained.Then,the distances between these pixels and central pixel were summed as their similarity weight.Finally,the capacity of capturing region details was enforced by the statistics within pixels of each local area and the central pixel in the degree of similarity,thus it could accomplish intensity inhomogeneity segmentation.Experimental results demonstrate that the improved model can segment intensity inhomogeneous images more precise than the Chan-Vese model and much faster than the Piesewise Smooth model.