Research of an improved dense matching algorithm based on graph cuts

According to the problem of graph cuts matching algorithm that lots of false matches appears in weak texture area, an improved graph cuts matching algorithm for dense matching is proposed in this paper. First, corner point detection and matching are made between stereo image pairs, and false matches are eliminated by relaxation algorithm and LMeds algorithm. Secondly, triangulate partitioning of discrete matching points is executed, the disparity of every triangle is control by its three vertex. Finally, the dense disparity mapping is draw by graph cuts based on energy function. The experimental results show performance of the proposed algorithm is improved and it will take a shorter time to compute an accuracy dense disparity mapping.