光学 精密工程, 2018, 26 (2): 435, 网络出版: 2018-03-21
检校场标定长焦航测相机参数
Calibration parameters of long focal aerial mapping camera by means of calibration field
长焦航测相机标定 检校场 相机参数 混合LMQN算法 focal aerial mapping camera calibration calibration field camera parameters Hybrid LMQN method
摘要
为实现长焦航测相机内参数及畸变参数标定,提出利用室外检校场对相机进行标定的方法。介绍了该方法由初值到精确值的两步法的求解步骤、计算过程及实验过程。首先通过给定已知特征点位置信息以及不同角度不同位置下拍摄的影像,通过特征点提取及匹配,根据像点坐标与世界坐标的线性对应关系,基于直接线性变换算法建立约束方程求取相机参数初值,进而通过混合LMQN(Levenberg-Marquardt与Qusai-Newton)迭代优化加速求解精确值。与精密测角法比无需精密设备的操作及记录,只需不同位置不同姿态拍摄多张包含精确位置信息的检校场影像即可。最后进行标定实验并对结果分析,基于检校场的标定方法在参数解算中特征点最大投影误差为2.471 pixel,标定参数主点精度为9.7 μm (<2 pixel),主距精度为4.3 μm (<1 pixel),解算相机拍摄位置精度0.035 m,满足测绘应用的精度需求。
Abstract
A two-step method was proposed to estimate parameters of long focal mapping camera by means of calibration field. The direct linear transform formulation(DLT) was used to obtain initial values for the hybrid LMQN iterative speed method by giving the feature points and the images taken from several positions and attitudes, and the whole process of calculating and experiment was described in detail. Compared with precise angle measurement method, operation and recording of accurate equipment are dispensable. It was only needed that images of calibration field with precise position information were taken from several different positions and orientations. The calibration results of experiment based on calibration field method indicate that the maximum re-projection error of feature points is 2.471 pixel, the calibration accuracies of principal point and principal focal distancereach 9.7 μm (<2 pixel) and 4.3 μm (<1 pixel), respectively, the calculated position accuracy of the camera is 0.035 m, which meets precision requirements of mapping application.
王欢, 周凌, 张鑫, 高思远, 初诞旭, 乔彦峰. 检校场标定长焦航测相机参数[J]. 光学 精密工程, 2018, 26(2): 435. WANG Huan, ZHOU Ling, ZHANG Xin, GAO Si-yuan, CHU Ting-xu, QIAO Yan-feng. Calibration parameters of long focal aerial mapping camera by means of calibration field[J]. Optics and Precision Engineering, 2018, 26(2): 435.