A new corner matching algorithm based on gradient

In this paper, a novel corner matching algorithm is presented. Gradient operator in Harris corner detector is directly used to compute the similarity between corners. Then initial matches are obtained. The algorithm is contrasted with normalized cross-correlation. Its computational complexity is reduced and matching speed is improved effectively because this method only adopts the addition and subtraction operations. Experiments on several real images test the matching speed, the matching precision and matching rate of the algorithm. The results demonstrate that the algorithm not only have higher speed but also get higher matching precision and correct matching rate. Even though the stereo image pairs have brightness differences, it still performs rather well.