提出一种基于亚像素精度的特征点提取算法,结合PnP方法和正交迭代(OI)算法计算乒乓球机器人本体的位姿.根据摄像机成像时弥散斑的近似高斯分布,以亚像素精度精确求取色标块的边缘,利用边缘直线交点得到高精度的角点作为特征点.基于PnP算法利用上述特征点求取机器人位姿的初值,再通过OI算法进行优化,以保证其姿态矩阵的正交性.实验结果表明,该方法能快速准确地实现机器人本体的位姿测量. A feature point extraction algorithm based on sub-pixel accuracy is proposed, which combines the PnP method with the Orthogonal Iteration (OI) algorithm to calculate the pose of the robot.According to the approximate Gaussian distribution of the speckle in the camera imaging, Take the edge of the color bar and use the intersection point of the edge to get the high-precision corner point as the feature point.Using the above feature points to get the initial value of robot pose based on the PnP algorithm, and then optimize by OI algorithm to ensure its attitude matrix Orthogonality.The experimental results show that this method can quickly and accurately measure the pose of the robot.