Multi-peak Candidate Matches Based Geometric Stereo Matching

Numerous algorithms for stereo matching have been developed and thousands of papers have been published over the years, some problems exist in stereo matching still. Aiming at some limits of existing algorithms, a novel stereo matching approach is developed based on multi-peak candidate sets combing with their geometric relationship. Gray cross correlation matching technique is adopted to extract multi-peak candidate correspondence sets. Correspondence is determined based on the local geometric relations between the three feature points and the multi-peak candidates. It could not be guaranteed that all valid correspondences were correct by local geometrical relational test alone. The disparity test and global consistency checkout are applied to eliminate the remaining ambiguous matches that are not removed by the local geometric relational test. Experimental results have shown that the proposed algorithm is feasible and accurate.