空间机械臂视觉相机内参标定技术研究

针对空间机械臂视觉相机的任务需求与配置，内参标定技术是相机高精度获取合作目标位置、方向等运动信息的首要前提和重要保障。文章设计了一种基于二维平面靶的机械臂相机内参标定方法，用于解算焦距、主点坐标、畸变系数等。首先，利用相机拍摄多幅平面靶图像，求解平面靶与相机图像之间的映射矩阵；其次，构建相机内参约束方程；再次，利用特征值分解，解算焦距、主点坐标等参数值；最后，引入径向、切向畸变构建非线性模型，将解得的内参作为初值，利用Levenberg-Marquardt优化算法即可计算出相机内参最优解。试验结果表明，该方法完全适用于机械臂相机内参标定，标定结果均满足机械臂视觉相机获取合作目标三维位姿的测量精度要求。

Research on Calibration of Intrinsic Parameters for Space Manipulator Camera Based on 2D Planar Pattern

Calibration of intrinsic parameters has been considered as an important premise and guarantee for estimating accurate position and orientation of a cooperative target to satisfying the task and configuration of space manipulator camera. A novel approach on calibration of intrinsic parameters for space manipulator camera based on a 2D target is designed in the article, which can be used to calculate a few parameters such as focal length, coordinates of principal point, coefficients of lens distortion. First of all, projection matrix be- tween 2D target and camera image can be acquired from several different target photos by camera; secondly, constraint equations of intrinsic parameters are formulated; thirdly, part parameters i.e. focal length and coor- dinates of principal point can be computed according to the eigenvalue decomposition （EVD）; finally, a non-linear model of camera is constructed by introducing radial and tangential distortion, and an optimal reso- lution of intrinsic parameters can be calculated by Levenberg-Marquardt algorithm based on considering above computed parameters as initial values. Experiment has shown that the approach fits for calibration of intrinsic parameters, and the calibration results all satisfy the required accuracy of 3D position and orientation of coop- erative target.