A progressive region-merging algorithm for stereo matching

A progressive region-merging algorithm for stereo matching is put forward in this paper. Firstly, the reference image is segmented by mean shift algorithm and the initial disparity maps are calculated using loopy belief propagation method. Secondly, we derive the reliable pixels through left-right consistent constraint. In addition, we present a new approach to exclude mismatch pixels in textureless regions, which will also reduce errors in the following region merging process. Finally, a progressive region-merging algorithm based on hierarchical clustering algorithm is put forward to merge neighboring regions and refine the plane parameters. Angle and distance are used together to measure whether the two planes are the same or not, which will reduce the calculation complexity. Experiments validate the proposed algorithm.