Reducing computation complexity for disparity matching

To facilitate the realization of free-viewpoint 3D video systems, disparity matching and view synthesis are two of the most significant operations. However, disparity matching demands high computation complexity which motivates the development of the proposed techniques. In this paper, we propose a shape-adaptive low-complexity (SALC) technique to remove computation redundancy between stereo image pairs for disparity matching. The novel idea takes advantage of that depth values of pixels inside the same object are either the same or gracefully changed, which implies that the operations of depth map generation may be reused, and are not necessarily computed pixel by pixel as conventional works did. Instead, the pixels with the same depth value should be treated as a group which becomes the basic unit in computing disparity matching. Meanwhile, the matching accuracy of stereo matching has been obviously improved by using searching blocks with shape information. From the experimental results, the proposed SALC technique accelerates the disparity matching for more than 26 times as well as improves quality of resulted depth maps with a 71.69% of bad pixel reduction compared with the conventional pixel-by-pixel disparity estimation.