移动机器人立体视觉高精度标定技术

A High-accuracy Calibration Technique of Stereo Vision for Mobile Robot

摄像机标定是移动机器人立体视觉的一个关键步骤,标定精度直接影响到障碍检测和路径规划的精度。在前人研究的基础上,提出了一种迭代方式的摄像机标定算法。算法将摄像机参数分为畸变参数和非畸变参数两类,每次迭代中固定一类参数来求解另一类参数,最终得到优化解。通过合理组织参数求解次序,迭代的每一步都可以通过最小二乘法得到闭式解,从而简化了计算。算法可以方便地进行扩展以包含不同类型的畸变参数,而不会增加算法的复杂度。实验结果表明此算法可以有效提高标定精度,可用于移动机器人的视觉系统。

Camera calibration is a key step in stereo vision for mobile robot. the calibration accuracy affects directly the precision of barrier detection and path planning. A camera calibration technique of iterative manner is presented on the basis of previous research. The camera parameters are divided into two categories. distortion and non-distortion parameters. fixed one type parameters in each iteration to solve the other parameters, obtaining the final optimal. Through the reasonable organization of the parameters order solving. every step of the iteration can acquire closed-form solution by least square method. which simplifies the calculation. The algorithm can be easily extended to include different types of distortion parameters. without increasing the complexity of the algorithm. Experiments validate the effectiveness of this technique in improving calibration precision.