基于平行直线束图像序列的摄像机标定

A Camera Calibration Technique Based on Images of Serial Parallel Lines Bundle

模拟与实现了一种基于主动视觉的摄像机参数标定方法;首先对空间平行直线束的成像规律进行研究,建立了平行直线束的消隐点与投影中心的连线和该平行直线束方向矢量的关系,在此基础上利用多组平行直线间的夹角作为约束求解摄像机焦距以及外参数。实验中采用普通的电脑液晶显示屏作为标定模板,液晶屏中显示按规律变换位置的平行直线束图像,摄像机进行同步拍摄获取液晶屏图像序列;在标定过程中摄像机与液晶屏的位置固定不动,避免了主动视觉方法对高精度平台的要求;该方法完全自动进行,内外参数分阶段线性求解且稳定性高;方法中不涉及图像间匹配问题,且无需知道标定模板的物理度量;模拟与真实实验均表明了该方法的有效性与实用性。

A new active vision based method for camera calibration is proposed. First of all, this paper researches imaging characteristics of parallel lines bundle, then sets up relationship between the connection line from vanishing point to projecting center and the direction vector of this parallel lines bundle. Based on above discuss, it uses angle among bundles of parallel lines as restriction to solve focus length and external parameters. Ordinary computer LCD screen is used as calibration plate in experimentation, images of serial parallel lines bundle whose position is changed by control are displayed on it, which are shot simultaneously by the camera to produce image series. During whole calibration process, the camera and LCD screen is kept stationary, thus avoids the use of high--precision drive platform required by other active vision based method. All the parameters can be solved linearly, stably and automatically. Image matching is no longer need to be handle and physical quantity is not need to be know. Simulator and real experiments demonstrate the effectively and practicability of our method.