摘要
针对数字工业摄影测量中的人工标志点的快速自动匹配问题,根据从物方空间角度描述核线的原理,提出了一种基于空间前方交会的匹配算法。该算法首先通过计算投影线间的最短距离来确定一组初始匹配像点;然后通过双片空间前方交会来确定一组潜在的物方点,并将其按距离分组;最后,将每一个潜在的物方点的坐标残差过大及属于同一张像片的像点剔除,同时通过判断与各物方点对应的像点数来确定同名像点。两组实验均证明,该算法不仅计算速度快,而且具有高匹配率和低误匹配率,能为数字工业摄影测量自检校光束法平差提供高精度的初值。
Aiming at achieving fast automatic matching of artificial target points in digital close range photogrammetry, a new matching method basing on two-image space intersection is presented in this paper, taking into account the epipolar line constraint in object space. Firstly, a set of initially matched image points is found by calibrating the shortest distance between two image rays. Secondly, the coordinate of corresponding object point is calculated through two-image space intersection, and these points are grouped according to the distances between each other. Finally, eliminate those image points with high rate of coordinate errors and which is in the same image, and find out homologous image points by numbers of image points corresponding with object points. Both experiences prove the advantage of the matching method, which is high speed, high matching quotient and low miss matching quotient. It can also provide precise initial values for the following bundle adjustment.
出处
《中国图象图形学报》
CSCD
北大核心
2009年第6期1194-1198,共5页
Journal of Image and Graphics
关键词
数字工业摄影测量
人工标志点
匹配
空间前方交会
digital industry photogrammetry, artificial target point, matching, space intersection
