A Region-Based Active Contour Model for Image Segmentation

The task of image segmentation is to partition an image into non-overlapping regions based on intensity or textural information. The active contour methods provide an effective way for segmentation, in which the boundaries of the objects are detected by evolving curves. In this paper, we propose a new region-based active contour model, which is based on the image global information for the stopping process. As a result, the model is robust to noise. Level set representation is used for the moving curves so that the topological changes during the evolution are handled automatically. Furthermore, an internal energy term is introduced, and it forces the level set function to be close to a signed distance function, which avoids the costly re-initialization for the evolving level set function. Experimental results demonstrate desirable performance of our model for images with large noise and complicated structures. Comparisons with Chan-Vese model and RSF model show the advantages of the model in terms of efficiency and accuracy.