期刊名称：International Journal of Multimedia and Ubiquitous Engineering
印刷版ISSN：1975-0080
出版年度：2015
卷号：10
期号：9
页码：137-146
DOI：10.14257/ijmue.2015.10.9.15
出版社：SERSC
摘要：Due to a high percentage of the bad pixels in small baseline stereo matching method, a small baseline stereo matching method based on adaptive weight is proposed. Firstly, the size of reference window is adaptively calculated for each reference point and then matching costs is evaluated according to adaptive weights and reference window size. Secondly, winner take all is used to evaluate initial disparities and unreliable matches are refused by using reliability constraints. Finally, disparity post processing method based on iterative diffuse using a new cost function is used to obtain a dense disparity map. A small baseline stereo image pair of Toulouse and stereo image pairs provided by the benchmark Middlebury database are used to test the proposed method. The experimental results show that this method can effectively reduce a percentage of the bad pixels of depth discontinuity areas, improve the overall disparity matching accuracy and meet the requirements of small baseline three-dimensional reconstruction.
关键词：stereo matching; small baseline; disparity; adaptive window; adaptive ; weight; disparity post processing
