针对空间交会对接的近距离位姿测量要求, 提出了一种基于单目视觉的二维合作目标位姿解算算法。为方便空中移动平台的调整以满足特定的位姿关系, 引入了一种新的姿态角定义方法, 此方法定义的三个姿态角可以作为平台姿态调整的反馈量且不受旋转顺序的限制。平面模型相对于相机坐标系的三个姿态角和位置向量可通过平面单应矩阵直接导出。在测量实验中, 算法基于DSP平台实现, 合作目标由4个共面LED光源构成, 测量值基准由高精度倾角传感器和全站仪获得。对空间位置变化范围为2 m×2 m, 姿态角变化范围为-30°~30°的目标平面进行测量, 结果表明, 本算法可实现0.88%的相对位置定位误差和最大为0.996°的姿态角测量误差, 且单帧算法的解算速度仅为0.25 ms。

A plane objects pose measurment algorithm based on monocular vision is proposed aiming at close-range pose measurement requirements in space rendezvous and docking. In order to facilitate the adjustment of the aerial mobile platform so as to satisfy the specific position relations, a new angle definition method is introduced, the three angles defined by which can be used as the feedback of the platform attitude adjustment and they are not restricted by the order of rotation. The three angles and position vectors of the plane model with respect to the camera coordinate system can be directly derived from the plane homography matrix. In the measurement experiment, the algorithm is implemented based on DSP platform, and the cooperative target is composed of 4 coplanar LED light sources, and the measuring datum is obtained by the high-precision inclination angle sensor and the total station instrument. Experimental results show that for the target surface measurement with spatial position change range at 2 m×2 m and the angle ranging from -30 degrees to 30 degrees, this algorithm can achieve 0.88% relative positioning error and maximum 0.996° angle measurement error, and the single-frame computing time is only 0.25 ms.