Automatic body segmentation with graph cut and self-adaptive initialization level set (SAILS)

With the extensive potential applications of computer technologies, automatic object segmentation plays a more and more important role in digital video processing, pattern recognition, and computer vision. In this paper, we propose an automatic human body segmentation system mainly consisting of human body detection and object segmentation. Firstly, an automatic human body detector is designed to provide hard constraints on the object and background for future segmentation. Secondly, in the first frame to be segmented, a coarse-to-fine segmentation strategy is employed to deal with the situation of partly detected object. By investigating the well-known graph cut segmentation algorithm and its implementations, we find that the segmentation error often occurs at the object boundary with the clattered background which will result in unpleasant visual artifacts in the segmented video. Therefore, background contrast removal (BCR) is proposed to weaken the high contrast in the background and preserve the contrast belonging to the foreground and background simultaneously. Thirdly, we propose a selfadaptive initialization level set (SAILS) to solve the tough problem, the undefined boundary in region-based segmentation, and to speed up the process of evolution simultaneously. Finally, an object updating scheme is proposed to detect and re-initialize new object when object disappears and reappears in the scene. Experimental results demonstrate that our body segmentation system works very well in the live video with strong edge and similar color in the background.