光学技术, 2019, 45 (4): 397, 网络出版: 2019-09-02
面向工业摄影测量的一种混合式相对定向算法
A hybrid algorithm of relative orientation in industrial photogrammetry
相对定向 工业摄影测量 本质矩阵 非线性优化 relative orientation industrial photogrammetry essential matrix non-linear optimization
摘要
针对工业摄影测量中多基线、大倾角交向摄影时相对定向困难的问题, 提出一种混合式相对定向算法。算法先利用本质矩阵分解确立相对定向参数初值, 再利用非线性优化计算精确解。从共面条件获得的行列式出发, 推导本质矩阵与相对定向参数的内在联系, 给出求解本质矩阵的两种方法。然后讨论本质矩阵奇异值分解计算旋转矩阵和平移矢量, 同时给出剔除伪解的依据。基于共面条件方程, 建立相对定向参数非线性优化的目标函数。定向精度对比试验验证了所提出的混合算法的稳定性和高精度, 对卫星天线曲面的测量试验也进一步说明该算法在工程实践中的适用性和可靠性。
Abstract
In order to satisfy the multiple-baselines and large-dip-angles photogrammetry applied in industry, a hybrid algorithm of relative orientation is presented. In this algorithm, the initial parameters of the relative orientation are established by the decomposition of the essential matrix, and then the exact solution of the parameters are calculated by a nonlinear optimization. First, the relationship between the essential matrix and the orientation parameters is derived by the determinant from the coplanarity condition, with two solving methods of the essential matrix being introduced, and then the rotation matrix and translation vector between two images are calculated by singular value decomposition(SVD) of the essential matrix, including the removal of the pseudo solutions. Second, according to the coplanarity condition equation, the cost function of relative orientation parameters for non-linear optimization is build. Finally, the result in comparison experiment indicates that the hybrid algorithm has acquired better stabilization and high precision, and the application in measuring curved surface of the antenna also identifies its practicability and reliability.
李云雷, 张曦, 屠大维. 面向工业摄影测量的一种混合式相对定向算法[J]. 光学技术, 2019, 45(4): 397. LI Yunlei, ZHANG Xi, TU Dawei. A hybrid algorithm of relative orientation in industrial photogrammetry[J]. Optical Technique, 2019, 45(4): 397.