中国空间科学技术

中国空间科学技术 ›› 2018, Vol. 38 ›› Issue (2): 1-8.doi: 10.16708/j.cnki.1000-758X.2018.0007

• 研究探讨 •    下一篇

一种基于基准优选的双目视觉位置解算方法

阳光烈，赵国伟*，罗焜   

  1. 北京航空航天大学宇航学院，北京  100083
  • 收稿日期:2017-06-30 修回日期:2017-08-30 接受日期:2018-01-15 出版日期:2018-03-25 发布日期:2018-04-25
  • 通讯作者: 赵国伟（1976-），男，教授，zhaoguowei@buaa.edu.cn，研究方向为航天飞行器设计、航天器动力学与控制等
  • 作者简介:阳光烈（1993-），男，硕士研究生，yangguanglie@buaa.edu.cn，研究方向为非合作目标位姿测量、航天器动力学与控制
  • 基金资助:
    国家自然科学基金（11572016）

A position calculating method of binocular vision based on datum optimization

YANG Guanglie, ZHAO Guowei*, LUO Kun   

  1. School of Astronautics, Beihang University, Beijing 100083, China
  • Received:2017-06-30 Revised:2017-08-30 Accepted:2018-01-15 Published:2018-03-25 Online:2018-04-25

PDF (PC)

34

摘要： 针对近距离空间非合作目标相对位姿测量任务需求，基于双目视觉原理建立了特征点位置测量模型与误差分析模型，分析了特征点位置解算奇异及奇异点附近位置解算误差过大的问题，提出了一种采用异面光轴配置和基准优选策略的双目视觉特征点位置解算方法。该方法通过错角安装相机光轴防止特征点在两个相机基准下进行位置解算时同时发生奇异，针对不同基准在测量域内误差分布的差异，建立了基准优选策略和相应的面向特定测量域的基准优选指标确定方法，从而获得了高精度测量结果。仿真结果表明，采用所提出的方法解决了特征点位置解算奇异问题，提高了测量域内特征点位置测量精度。

关键词: 非合作目标, 双目视觉, 奇异, 测量基准, 位置测量

Abstract: Aiming to the requirement of relative attitude and position measurement in near-field for space non-cooperative target, position measurement model and error analysis model of characteristic point were established based on the theory of binocular vision. The problem of error increasing due to calculation singularity was analyzedas well. Based on aproposed optimization strategy, a characteristic point calculation method was founded for binocular vision system with non-coplanar optical axes. This method avoids the simultaneous singularity under both cameral datum by configuring the optical axes with certain alternate angle. According to the different measrement error distribution of each datum, the datum optimization strategy was established and the parameter determining method of corresponding criterion was proposed to obtain high precision measurement results. The numerical simulation result shows that, the proposed method can solve the calculation singularity problem and improve the precision of characteristic point position measurement in specific measurement domain.

Key words: non-cooperative target, binocular vision, singularity, measuring datum, position measurement