基于几何不变量的直线匹配方法

摘要

针对目前直线匹配算法大多基于直线段支持区域的几何特性、灰度特性,但这些特性又极易受外界因素干扰造成匹配失败的情况,提出了基于直线几何不变量的匹配方法,该方法在无任何已知条件的情况下,运用叉积符号、点的线束交比不变量进行直线匹配.对于各个视点图像,采用Hough变换提取直线间交点集,再采用Graham扫描法构造一个最小的凸包,并通过匹配凸包顶点的交比序列来建立点与点的对应关系,选取其中一个对应关系完成直线的匹配.实验结果表明,该方法鲁棒性较好,具有广泛的应用性.%Aimed at features of the current Line matching algorithm, such as geometric properties and gray properties of support region of straight line are used to match straight line. But they are easy to be interfered by the environment and lead to failure of match. A method for matching straight line based on geometric invariance is proposed. Without any constraint and the given conditions, straight Line based on invariant of cross product symbol and harness cross ratio can be matched perfectly. For each view images, firstly, crossing point set of lines are extracted by Hough transform, the least convex hull is constructed by Graham scan method. Then, the correspondence relationship between points is set up by matching sequences of vertex cross ratio of convex hull, finally, matching is implemented via one of the correspondence relationship is selected. Experiment results show that the method has better robustness and has a wide range of application.