机器人单目视觉标定圆心靶标特征点几何求解

Robot Monocular Vision Calibration Center Target Feature Points Geometric Calculation

针对机器人单目视觉标定圆心靶标特征点求取问题,利用射影变化公切线不变性,建立圆心靶标的特征点几何求解模型,并进行了精确求解,该模型理论上不存在模型误差;为了检验几何求解模型的精确性和稳定性,分别对机器人单目视觉坐标系下的圆心靶标圆心点和棋盘格靶标角点进行了三维测量。实验结果表明:测量长度在300 mm内的最大相对误差小于1.5%,测量结果较精确;各点的距离平均绝对偏差均在0.2 mm左右,且无阶跃,该方法稳定。

Aimed at the problems of feature point solution of robot monocular vision calibration center target, projective change tangent invariability was used to establish the model of geometric solution of feature point of center target and to solve accurately. There were no modeling errors in the model in theory. In order to test precision and stability of geometric solution model, measure in three di- mensional （3D） was carried out in part for center target feature points under the coordinate system of robot monocular vision calibration and for the plane grid target corner points. Experimental results show that the maximum relative error is less than 1.5% for the measuring length within 300 mm, and precision is quite high. The average absolute difference is all at about 0.2 mm for the distance between each point with no step jump, and the method is stable.