Image Segmentation Algorithm Based On Chan-Vese Model

Compared to traditional image segmentation, PDE method involves too many calculations. But this flexible numerical method has good stability when discreting PDE. Besides, it can carry out high-quality image restoration and precise image segmentation. Thus, this method gains more and more attentations in image processing.In this article, we firstly give a comprehensive summary of present image segmentation methods. Then, after introducing active contours model and level set method, we propose two new algorithm methods by analysing Chan-Vese model: (I) Fast adaptive image segmentation algorithm based on Chan-Vese model. By studying the relation of the initial level set location and the segmentation speed, we indicate that the initial location can affect the speed of the curve evolvement and the level set evolving speed can be increased by optimizing the initial location; (II) Medicine image segmentation algorithm based on OCV model. First, we partition the image into four parts: target, background, edge and noise by Otsu segmentaion algorithm. Then by energy function, we judge whether the pixels belong to the background or to the target, and fine-tune the initial segmentation result by using the global information of homogeneous region to obtain more precise segmentation result.Finally, we point out the direction of further research.