一种山体图像边缘快速立体匹配方法

A Fast Stereo Edge-Matching Algorithm Based on Mountain Images

该文提出了一种新的针对山体图像边缘的立体匹配算法。为了保证匹配的准确性,首先引入极线方向上的视差梯度约束,然后在考察唯一性的基础上,通过搜索并考察边缘端点周围的角点信息来引导边缘的匹配,并且限定边缘端点的角点搜索范围,最终匹配结果保证了两幅图像中的匹配对是一对一的唯一对应。实验结果表明,不同约束方法的融合在很大程度上提高了匹配速度和匹配精度。

A new fast stereo edge-matching algorithm based on mountain images is presented. In order to ensure the accuracy of the match. Firstly the disparity gradient restriction across the direction of epipolar lines is used, and then on the basis of exclusionism searched the corner information close to edge point to guide the matching with restricted corners-searched range. The final match results ensure that the matching pairs in two images are one-to-one correspondences. The algorithm results show that the amalgamation of different restrictions improved the matching speed and precision.