Study On Chan-Vese Active Contour Model

Image segmentation is a basic and important issue in image analysis and computer vision. The goal of image segmentation is to divide an image into disjoint subdomains, each of which is gray-homogeneous. There are many methods of image segmentation. Partial differential equations based active contour model segmentation method is one of the most successful methods for image segmentation.The model researched in this paper is a famous region-based geometric active contour model. Because it has strong applicability for change of topologic structure of the target object, it is used widely and has many following research. This model resumes image made up by two homogeneous region (object and background) which have different means, so we can use in the difference of average grey between object and background to divide image. Its essence is to use binary step constant function to approximate the image to be segmented. Power-law transformation and negative transformation are two basic types of gray level transformations that are used frequently for image enhancement. This paper incorporates the two transformations into the well-known Chan-Vese model to raise the performance of Chan-Vese model in terms of segmentation speed and quality. By practical experiments, it is verified that new Chan-Vese model has faster convergence than Chan-Vese model, and exhibits certain capability of handling straight edges and corners.