基于位姿约束的大视场双目视觉标定算法  

Large Field and Binocular Vision Calibration Algorithm Based on Position and Orientation Constraints

作　　者：张超[1] 韩成[1] 杨华民[1] 杨帆[1] 

机构地区：[1]长春理工大学计算机科学技术学院,吉林长春130000

出　　处：《光学学报》2016年第1期197-206,共10页Acta Optica Sinica

基　　金：国家科技支撑计划重大项目课题(2012BAF12B22);吉林省重大科技攻关项目(2012ZDGG004);吉林省科技发展计划项目(20130303011GX);吉林省科技发展计划项目(20140204050GX);长春市科技计划项目(14KG013)

摘　　要：为了提高大视场、远距离的双目摄像机标定精度,提出一种基于位姿约束的摄像机标定算法。该方法利用双目摄像机之间的三维位姿关系是刚体变换这一属性,标定出左、右摄像机相对位姿的外部参数。利用相对位姿为约束条件求取摄像机的初始内部参数,剔除较大的重投影误差值对应的标定图像组,重复迭代直至重投影误差平均值小于指定值,得到多个待优化的摄像机内部参数。再将最后标定图像组的角点坐标、待优化的摄像机内部参数和相应的外部参数,建立一个以角点三维重构坐标值与实际设定角点三维坐标值的模均值为最小的目标函数,求解出双目摄像机标定参数的最优解。该方法很好地解决了误差大的标定图像造成的影响,且充分利用了双目摄像机之间的位姿约束关系。通过仿真和标定实验可以看出,本文方法可以实现大视场双目摄像机的高精度标定。In order to improve the accuracy of large field, far distant binocular camera calibration, a new calibration algorithm based on the constraints of camera position and orientation is proposed. As the three-dimensional position and orientation relationship between binocular cameras is a rigid transformation, the external parameters of relative pose of left and right cameras are targeted. By using the constraints of the relative position and orientation, internal parameters of the initial left and right cameras are calibrated. Excluding calibration image groups by large re- projection errors, repeating iterations until the mean of re-projection errors is less than the specified value, sev^eral camera internal parameters to be optimized is obtained. Then combIning with the corner point coordinates／of image groups, camera internal parameters to be optimized and corresponding external parameters, an objective function is established that the values of three-dimensional reconstruct corner point coordinates and the values of the actual three-dimensional corner point coordinates are the smallest. The optimal solution of binocular camera calibration parameters is solved. The proposed method is a good solution to the impact caused by the large error of calibration image, and makes full use of the constraint relationship between the binocular cameras position and orientation. Through simulation and calibration experiments, it can be seen that the proposed method can achieve a high precision calibration of large field binocular cameras.

关 键 词：机器视觉 双目视觉 位姿约束 相机标定 刚体变换 重投影误差 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]