摄像机全自动标定方法研究  

Research on Fully Automatic Camera Calibration Method

作　　者：许勇[1,2] 郭鹏宇[1,2] 龙古灿[1,2] 张小虎[1,2] 于起峰[1,2] 

机构地区：[1]国防科学技术大学航天科学与工程学院,湖南长沙410073 [2]国防科学技术大学图像测量与视觉导航湖南省重点实验室,湖南长沙410073

出　　处：《光学学报》2013年第6期131-141,共11页Acta Optica Sinica

基　　金：国家自然科学基金(11072263;11272347)资助课题

摘　　要：提出了一种基于空间控制点的全自动标定方法,该方法只需知道控制点的空间坐标及其图像,通过自动提点、自动匹配以及高精度解算三个过程完成全自动标定。针对已知光心位置和已知光轴指向两种情况,对摄像机标定进行了理论推导,提出了焦距一致判据的自动匹配算法;针对不存在先验信息的情况,以计算残差和重投影残差为判据,根据最近邻原则提出了残差最小自动匹配算法。仿真和实验验证了提出的自动匹配算法和全自动标定模型的快速、正确性和高精度。Attention is focused on fully automatic camera calibration, and a calibration method based on three-dimensional （3D） reference points is proposed. This method acquires camera parameters through three steps： automatic recognition, automatic matching and high-precision calculation, which only needs the coordinates of 3D points and the image. The method of calibration and the focal length coherence automatic matching algorithm are put forward under the condition of knowing the coordinates of camera center and the orientation of camera optical axis, respectively; and the minimal residual automatic matching algorithm is proposed using computation residual and reprojection residual as criterion, when there is no prior information. Numerical simulation and physical experimental results show the speediness, correctness and high precision of our automatic matching algorithms and fully automatic calibration model.

关 键 词：测量 机器视觉 全自动标定 残差最小自动匹配 焦距一致自动匹配 摄像测量 

分 类 号：O436[机械工程—光学工程]