基于固定视点的视觉标定算法研究

Camera calibration algorithm based on stationary view

针对在固定视点中机器视觉标定过程复杂、效率低、标定精度不高的问题,对目前普遍应用的摄像机标定和手眼标定算法进行了研究,提出了可一次性完成固定视点的视觉标定算法。利用实验室的MOTOMAN-UP6工业机器人对该标定算法进行了验证,首先由机器人末端夹持一个棋盘格完成一系列的运动,每次运动结束时摄像机拍摄一幅棋盘格图像,并记录下机器人末端的位姿,等所有运动完成后,采用张正友标定算法对摄像机进行标定,接着采用基于最小二乘法的标定算法对摄像机与基座关系进行简便高效的标定。研究结果表明:该算法标定过程简单、求解速度快并且结果精度高,具有很强的实用性。

Aiming at the problem of complex process,low efficiency and low accuracy in camera calibration,the widely used camera calibration and hand-eye calibration algorithm was studied,and a new camera calibration algorithm based on stationary view that can be completed only once was proposed. This calibration algorithm was tested by MOTOMAN-UP6 industrial robots in laboratory. Firstly,a chess board was holded by robot manipulator and a series of movements was completed,and in the end of each movement,one chess board picture was grabbed by the camera and the pose of the manipulator was recorded. After all movements were completed,the camera calibration was done by Zhang＇s algorithm and the relationship calibration between the camera and the robot base was done by the calibration algorithm base on least square method. The procedure is simple and efficient. The results indicate that this algorithm operates simply,solves quickly and precisely,and has strong practicability.