近景摄影测量系统中非编码点匹配算法研究

从同一场景的多幅图像中找出对应的非编码点,是实现近景摄影测量的必要前提。分析了同场景多幅图像中非编码点之间的几何约束,提出了以三视极线和图像拓扑结构为约束的离散图像点匹配算法。先将同时满足三视图三极线和图像拓扑结构约束的点作为确定的匹配点并赋予编码值;再利用已经编码的标记点,以对极几何和图像拓扑结构为约束对剩下的未匹配点进行匹配。将该方法应用于铁牛三维光学测量系统的特征点测量系统进行的实例验证表明,当拍摄像平面和物平面夹角较小时,匹配完全正确;而在一般的拍摄条件下,其正确匹配率约为97%。

Research on Matching Methods of Non-Coded Points in Close Range Photogrammetry

Finding the corresponding non-coded points among multi-images of the same scene is a premise of close range photogrammetry.The geometrical constraints among multi-images of the same scene are analyzed and a matching method among discrete non-coded points is proposed based on constraints of tri-view polar and image topological structure.Points that simultaneously satisfy constraints of both tri-view polar and image topological structure are matched and coded first and the other unmatched ones are coded by using the coded ones and constraints of epipolar geometry and image topological structure.The matching method has been applied to the 3D Optical Measurement System of Feature Points（TN 3DOMS.FP） developed by Saint Buffalo Technologies Ltd.Company and quantity measurement instances show perfect matching when the angle between the graphing plane and the graphed object plane is small and its general correct matching ratio is around 97%.