Abstract: A level set image segmentation model presented by ChanVese does not rely on the image edge information and it has good segmentation effect on weak edge and the images containing internal outline. However, for the gray scale of background image contains two or more hierarchies, image segmentation can not get accurate results. This paper puts forward a new improved algorithm based on CV model. The algorithm introduces an improved fast CV method into the global gradient information and the prior knowledge of the target. The experimental results show that the method can identify the complex grayscale background image with target area profile of multiple hierarchies, which proves to have favorable performance and adaptability.

Key words: image segmentation; CV level set; fast; prior knowledge; background gray multi-level hierarchy