摄像机参数与地面点坐标改正数的分离估计

摘要

In the geo-location method based on common points in multiple aerial images, the computational efficiency of the simultaneous least square estimate used to reckon the corrections of aerial camera parameters and ground point coordinates degrades rapidly with the increase of the number of common points. To solve this problem, a novel algorithm is presented to separately estimate the corrections of camera parameters and ground point coordinates by considering the feature that the coefficient matrices of the corrections of camera parameters and ground point coordinates are independent. It is theoretically based on the separate least square estimation, and calculates the corrections of camera parameters and ground point coordinates in every iteration step alternately. In the iteration procedure, the coefficient matrices are parted to decrease their dimension numbers. Such partition improves the efficiency of correction computation. Simulation results show that the separate algorithm has higher computational efficiency than the present simultaneous algorithm, without degrading the precision of the corrections.%在基于多帧航空图像同名点的对地定位方法中,求解摄像机参数改正数和地面点坐标改正数的一次完成最小二乘算法(LMS)的运算效率随同名点数目的增加而急剧下降.为解决该问题,根据改正数约束方程中摄像机参数改正数系数矩阵与地面点坐标改正数系数矩阵相互独立的特点,提出了摄像机参数改正数和地面点坐标改正数的分离估计新算法.该算法以分离LMS为基础,用迭代方法交替求解摄像机参数改正数和地面点坐标改正数.每一步迭代计算均按分块方式处理系数矩阵,降低了系数矩阵维数,为改正数解算效率的提高创造了条件.仿真分析结果表明,在保证改正数计算精度的同时,分离估计算法的运算效率优于一次完成算法.