Improved Adaptive Support-weight Algorithm Based on Guide filtering and Disparity calibration
Xiao JingWen, Min Wan, Wu Yao · Advances in computer science research · 2015
A novel local stereo matching method based on guided filtering and disparity calibration was presented in this paper.Firstly, the three dimensional cost volume was built by computing initial matching costs at given disparity values, and it was enhanced with guided filtering to preserve edge information effectively and improve operational efficiency; secondly, a disparity calibration approach was designed to refine the initial disparity, so the matching accuracy can be further improved.Experimental results on Middlebury dataset demonstrate that this proposed technique can obtain accurate and dense disparity map while eliminating effectively matching ambiguity, without affecting the running speed.