改进了用于标定线阵摄像机的传统精密测角算法, 标定用于面阵摄像机的参数。该算法利用两束平行光之间的夹角和投影在摄像机上图像点之间的对应关系, 在给定一个预测摄像机主点的基础上计算它和实际主点之间的偏差以及摄像机焦距。分析了图像特征提取误差对于平行光夹角测量精度的影响, 并给出一种基于平行光夹角误差最小的最优估计, 从而进一步提高摄像机内部参数的标定精度。通过仿真实验分析了图像特征提取精度和平行光夹角测量精度对摄像机参数标定精度的影响。结果显示, 当图像特征提取精度为0.1 pixel, 二维转台精度为0.5″时, 主点标定精度可以达到0.56 pixel, 焦距标定精度可以达到0.06 mm。利用精度为0.5″的二维转台对摄像机参数进行了实际标定, 通过分析像点和标定结果所计算的平行光夹角和实际测量的平行光夹角的误差, 可知本文算法的误差是经典精密测角法的686%, 由此证明该算法对于面阵摄像机参数标定具有更好的结果。

An improved exact measuring angle algorithm is proposed for parameter calibration of area array cameras. Based on a given estimated principal point of camera, the algorithm calculated deviation between the estimated and actual principal points and focal distance of the camera by using the corresponding relationship between intersection angle of two beams of parallel lights and image point of projection on the camera. The influence of image feature extraction error on intersection angle measurement accuracy of parallel lights was analyzed and an optimal estimation based on the minimum intersection angle error of parallel lights was employed so as to promote the calibration accuracy of internal parameter of the camera. Simulation experiment was conducted to analyze the influence of image feature extraction accuracy and intersection angle measurement accuracy of parallel lights on camera parameter calibration precision. The results show that the calibration accuracies of principal point and focal distance can reach 0.56 pixel and 0.06 mm respectively when image characteristics extraction accuracy is 0.1 pixel and two-dimensional rotary table accuracy is 0.5″. An two-dimensional rotary table with accuracy reaching 0.5″ is used for parameter calibration of the camera. In comparison of intersection angle error of parallel light calculated based on image point and calibration results with the calibration result, the error of the improved algorithm is 68.6% of the classic exact measuring angle method. This proves that the proposed algorithm has a better accuracy in parameter calibration of area array camera.