An improved minimum spanning tree stereo matching algorithm

摘要

The minimum spanning tree stereo matching algorithm only takes one channel of R, G, B channels into account, ignoring the effect of the other two channels on the final edge weight in color image. This paper proposes an improved minimum spanning tree stereo matching algorithm, which calculates the weighted euclidean distance using a three-channel approach, and combines three-channel edge weight. This algorithm has been tested on Tsukuba, Venus, Teddy, Cones image. Simulation results show that the improved algorithm enhances the robustness of edge weight function and the connectivity of minimum spanning tree. It not only improves stereo matching accuracy, but also increases computational speed.