基于双目立体视觉的月球车三维定位算法

摘要

In order to satisfy the requirement of autonomous navigation of lunar rover, a three-dimensional localization algorithm of lunar rover based on binocular vision was presented in the paper. Firstly, the three-dimensional localization system frame was proposed, and the important modules were analyzed. Then, such image processing as image preprocessing, feature extraction, stereo matching, 3D reconstruction, feature tracking and matching method were studied to establish accurate environment features correspondence relations. On this basis, a position and orientation change estima tion algorithm of lunar rover was described. Finally, an experiment was designed to confirm the effectiveness of the proposed method.%为了满足月球车自主导航的需要,提出了1种基于双目立体视觉的月球车三维定位算法.建立了基于双目立体视觉的月球车三维定位系统框架图,并对框架中主要模块进行了分析；研究了图像预处理、特征点提取、特征点的立体匹配、三维重建和特征点的跟踪匹配算法等图像处理环节,以获得月球车运动前后相匹配的环境特征点,并提出了1种鲁棒的月球车位姿变化估计算法.通过实际场地试验验证了算法的有效性.