为了通过视觉测量的方法实现小型空间机器人协同位姿的求解, 设计了基于单目视觉的协同位姿测量系统。针对复杂的太空工作条件, 设计了作用范围广、精度高且稳健性强的合作靶标; 设计了利用图像梯度法提取单像素边缘的算法, 并通过中心匹配等约束完成了靶标识别; 推导了特征点距离与位姿测量精度的关系, 在此基础上提出了基于分区处理的特征点提取策略, 实现了复杂条件下特征点的分区提取和准确编号, 并扩展了工作距离, 改善了位姿求解精度。实验结果表明, 针对640 pixel×480 pixel图片, 在Matlab环境下处理速度约为84.3 ms/frame, 在2.2 m距离范围内位姿测量误差在5 mm、1°以内。设计的单目视觉位姿测量系统满足小型空间机器人协同工作的任务需求。

In order to solve the cooperative pose of small-scale space robots by means of visual measurement, a cooperative pose measurement system based on monocular vision is designed. Firstly, a cooperative target with wide range of functions, high precision and strong robustness is designed for complex space working conditions. Secondly, one-pixel-width edge extraction algorithm is designed by image gradient method, and target identification is done with center matching and other constraints. Thirdly, the relationships between distance of feature points and precision of pose measurement are derived, and the extraction strategy of feature points based on partition processing is designed to realize the extraction and number of feature points in complex conditions, expand the working distance and improve the precision of pose measurement. The experimental results show that the processing speed is about 84.3 ms/frame in Matlab for the image of 640 pixel×480 pixel. Pose measurement errors are less than 5 mm and 1° within the range of 2.2 m. This system meets the cooperative task need of small-scale space robots.