物方面元匹配在倾斜摄影三维重建中的应用

Application of Object Patch Matching in 3D Reconstruction of Oblique Photogrammetry

采用传统的最小二乘影像匹配方法匹配倾斜影像时,倾斜影像灰度和几何畸变较大,地物遮挡较多,导致匹配到的同名像点分布不均匀,匹配可靠性较低影响三维重建效果。提出采用物方面元匹配方法对倾斜影像做精化匹配。匹配方法的过程是:首先根据初匹配生成的物点信息,采用物方差分法计算物方面元的法向量方向角;然后根据极限约束估计同名像点坐标,进而前方交会求解物方面元高程初值;最后将物方面元法向量方向角和高程初值带入条件方程,做最小二乘匹配迭代计算,实现倾斜影像同名点匹配。在经过实验验证后发现:本文方法用于倾斜影像匹配时,可提高同名点匹配的可靠性和分布的均匀性。

In order to solve the problem of uneven distribution of correspondence points and low matching reliability,the traditional least square image matching method was introducedto make precise matching of oblique image caused by the large gray and geometric distortion of oblique image and the more occlusion of ground objects.The process of the matching method in this paper was:firstly,according to the information of the object points generated by the initial matching,the normals direction angle of the object patch was calculated using the object variance method,the coordinates of correspondence points wereestimated according to the limit constraint,and the initial elevation value of the object patch was solved by forward intersection.Finally,the direction angle of the normals and the initial elevation value of the object patch werebrought into the condition equation to do the least square matching iterative calculation,The correspondence points matching of oblique image was realized.After experimental verification,it was found that this method couldimprove the reliability and distribution of correspondence points matching.