空间非合作目标的三目立体视觉测量方法  

Trinocular Stereo Visual Measurement Method for Spatial Non-Cooperative Target

作　　者：束安 裴浩东[1,3] 段慧仙[3,4] Shu An;Pei Haodong;Duan Huixian(Key Laboratory of Intelligent Infrared Perception,Chinese Academy of Sciences,Shanghai 200083,China;University of Chinese Academy of Sciences,Beijing 100049,China;Shanghai Institute of Techmical Physics,Chinese Academy of Sciences,Shanghai 200083,China;Key Laboratory of Infrared System Detection and Imaging Technology,Chinese Academy of Sciences,Shanghai 200083,China)

机构地区：[1]中国科学院智能红外感知重点实验室,上海200083 [2]中国科学院大学,北京100049 [3]中国科学院上海技术物理研究所,上海200083 [4]中国科学院红外探测与成像技术重点实验室,上海200083

出　　处：《光学学报》2021年第6期157-165,共9页Acta Optica Sinica

基　　金：“十三五”国家重点研发计划高光谱红外基准载荷技术研究(Y8K22017N1)。

摘　　要：针对在轨服务任务如何在近距离处获取空间非合作目标的相对位置和姿态的难题,提出一种三目立体视觉测量方法。首先利用三台呈等边三角形布置的大视场可见光相机获取图像;然后采用所提方法对特征点进行匹配;接着采用RANSAC方法解算被测目标在世界坐标系下的位姿参数;最后通过对非合作卫星模型的静止、位置移动和姿态转动进行实验。实验结果表明,静止状态下的相对位置精度优于2.2 mm,相对角度测量精度优于0.3°;当模型位置移动时,绝对位置精度优于3 mm;当模型姿态转动时,相对位置精度优于5.6 mm,相对角度测量精度优于1.7°,说明所提方法可以改善双目立体视觉技术在测量角度大的区域易出现测量盲区、特征点定位误匹配和测量视场有限的不足。Aiming at the problem of how to obtain the relative position and posture of non-cooperative targets in space at close range for on-orbit service mission, a trinocular stereo vision measurement method is proposed. First, three large-field visible light cameras arranged in equilateral triangles are used to acquire images. Then, the proposed method is used to match the feature points. Second, the RANSAC method is used to calculate the pose parameters of the measured target in the world coordinate system. Finally, the static motion, position movement and attitude rotation of the non-cooperative satellite model are tested. Experimental results show that the relative position accuracy is better than 2.2 mm in the static state, and the relative angle measurement accuracy is better than 0.3°;when the model position moves, the absolute position accuracy is better than 3 mm;when the model posture rotates, the relative position accuracy is better than 5.6 mm, the relative angle measurement accuracy is better than 1.7°, indicating that the proposed method can improve the shortcomings of binocular stereo vision technology, such as measurement blind spots, mismatched feature points location and limited measurement field of view.

关 键 词：机器视觉 三目视觉 ORB 非合作目标 极线约束 随机抽样一致性 

分 类 号：TP391[自动化与计算机技术—计算机应用技术] V557[自动化与计算机技术—计算机科学与技术]