Adaptive Superpixel-Based Disparity Estimation Algorithm Using Plane Information and Disparity Refining Mechanism in Stereo Matching

The motivation of this paper is to address the limitations conventional keypoint-based disparity estimation methods. Conventionally, usually based on local information keypoints. However, keypoints may distribute sparsely in smooth region, and with same descriptors appear a symmetric pattern. Therefore, methods have limited performance regions. proposed algorithm superpixel-based. Instead performing keypoint matching, both semiglobal are applied determine algorithm. Since semi-global superpixel applied, accuracy can be improved, especially for Moreover, non-uniform distribution problem keypoints, refining mechanism similarity distance neighboring superpixels correct no or few experiments show that map generated by has lower matching error rate than other