大场景深度范围下的角度校验立体匹配算法

Stereo Matching Algorithm Based on Angle Check for Large Scene Depth Range

针对三维扫描系统开发过程中遇到的大场景深度范围下标志点误匹配率较高的问题,对基于单应性约束的双目立体匹配算法进行研究,从理论上推导出匹配误差与场景深度和工作距离比值间的数学关系,明确该算法的适用条件。为提高场景深度与工作距离比值较大时的匹配正确率,提出一种基于角度校验的双目立体匹配算法,首先通过极线约束获得初始匹配点对,然后通过计算角差异度校验得到最终正确匹配点对,实验表明所提出的方法能够达到良好的匹配效果。

Aiming at the problem of marker mismatching increasing in large scene depth range during the development of three-dimensional scanning system,the stereo matching algorithm based on homography constraint is studied.The mathematical relationship between matching error and the ratio of scene depth to working distance is deduced theoretically,and the applicable conditions of the algorithm are clarified.In order to improve the matching accuracy when the ratio of scene depth to working distance is large,a stereo matching algorithm based on angle checking is proposed.First,the initial matching point pairs are obtained by epipolar constraint,and then the correct final matching point pairs are obtained by angle difference checking.Experiments show that the proposed method can achieve better matching effect.