A new stereo matching algorithm based on image segmentation

Comparing with the local algorithm in stereo matching, the global algorithm can get highly quality disparity space image (DSI) usually. The graph cuts algorithm, as the traditional global matching method, can get the highly quality DSI with long computation time, so the graph cuts algorithm is difficult to used in real-time matching applications. In this paper, a new method is proposed to accelerate the matching speed. Firstly, the segments of the image are calculated using the Gaussians pyramid (DoGs) computing method, secondly calculate the disparity range of two corresponding image segments in image pair using the color feature, and every segment pair is matched based on graph cuts algorithm according to the offset of disparity range in parallel time, finally all the segments are assembled into a whole disparity image. The experimental results show that the matching speed is accelerated greatly with a high disparity image quality.