Vertical Edge Depth Estimation

Through binocular vision the fast disparity image is often desired for many applications, but, stereo vision still is an on studying subject. It is hard to get the precisely result because of the complex of the vision task. Vertical edges of depth images are presented much information of the stereo scene for further image processing. We present an image processing technique that can fast estimate vertical edges depth from binocular vision images. By finding out meaningful dots which present the best matched edges in the disparity space image, the depth can be estimated. When detecting these dots, a cross-emphasizing filter is used. The final depth estimation will be presented after belief propagation.Our method is a compromise between local methods and global optimization.