基于纯平移两视图几何的镜头畸变参数标定

Calibration of lens distortion parameters based on two view geometry of translation motion

针对摄像机镜头畸变对系统测量精度的影响,提出了基于纯平移两视图几何的镜头畸变参数标定方法。首先,分析了机器视觉测量中影响测量精度的主要畸变类型,建立了镜头非线性畸变模型。然后,利用射影几何及纯平移两视图几何的固有特性,构建了四组约束方程用于求解畸变参数。最后,针对大视场测量时标定板无法有效充满视场的问题,提出了利用四维电控平台对视场分区域拍摄多组纯平移运动图像的畸变参数标定流程,利用较小的靶标实现了大视场的镜头畸变参数标定。在实验室验证了本文提出的畸变校正方法的可行性。结果显示:提出的方法标定精度较高,标定后图像特征点连线的直线度误差减小了89%,标定精度及可靠性均满足机器视觉测量的要求。

In consideration of the effect of lens distortion on the measuring accuracy of a camera,a lens distortion calibration method based on two view geometry of a translation motion was presented in this paper.The main types of lens distortion affecting the measurement accuracy were analyzed,and a nonlinear distortion model was established.Then,four group equations were established for solving distortion parameters by using the inherent characteristics of projective and two view geometries.Finally,to solve the problem that the small pattern can not effectively be full of the field of view（FOV）in a large range measurement,a calibration processing of distortion parameters by acquiring multigroups＇ translation motion images in sub-regional FOV was put forward,by which the calibration of the distortion parameters was implemented in a large FOV by the small pattern.The feasibility of the calibration method was verified by a library,and the results show that the calibration precision of proposed method is higher and the straightness error has reduced by 89%.The calibration accuracy and reliability of this method meet the requirements of visual measurement.