Integration of optical flow and dynamic programming for stereo matching

Abstract

The authors present a novel approach to integrate the optical flow estimation into a block stereo matching algorithm using dynamic programming (DP). It is well known that reliable ground control points can significantly improve the performance of stereo matching, but false matches can also significantly degrade the performance of stereo matching. In order to extract as many reliable ground control points as possible, the authors use the Lucas–Kanade method in pyramids to estimate the optical flow of every pixel and use a bidirectional matching process to remove false matches. To further reduce the impact of false matches, a matching cost is assigned to each point to form a set of ‘soft’ ground control points. Experimental results show that the proposed algorithm, which uses ‘soft’ ground control points and DP, can achieve the performance close to the best performance of algorithms using DP.