Study of occlusions problem in stereo vision

This paper presents a symmetric multiple windows using improved fast algorithm in stereo matching. The symmetric multiple windows method is an adaptive and stable scheme, which can smooth depth discontinous area. The method of single matching phase is used to estimate disparity. The adaptive property of symmetric multiple windows are adopted to reduce mistake matching probability. To produce smooth and detailed disparity maps, two assumptions that were originally proposed by Marr and Poggio are adopted: uniqueness and continuity. The improved sum of squared differences (SSD) algorithm avoids repeatedly computing, reduces the computing complexity efficiently. This paper demonstrates and discusses performances with real stereo pairs, the algorithmpsilas excellence can efficiently identify and reduce mistake matching in occluded region, then produce smooth image, and shorten matching time.