摘要
针对空间目标捕获任务技术需求,设计了一种基于平面靶标特征点提取的单目视觉线性相对位姿解算方法.提出基于几何指数阈值判别的特征光标提取方法,并采用非线性项变量代换方法,设计了基于特征光标提取的线性相对位姿解算方法,建立了单目视觉相对测量地面实验系统,在模拟目标航天器相对位置和姿态低速动态漂移的条件下进行了试验.测量结果表明,所提出的特征光标提取算法能够有效避免高亮光斑对特征光标提取的不利影响,且相对位姿解算算法具有较好的精度和实时性,相对于传统非线性迭代求解方法,该方法计算量小,求解过程稳定.
A linear monocular vision determination method is proposed based on feature point extraction for orbit target capture. With a consideration of the interference from the highlight facula, the feature point extraction method based on geometry index threshold is designed, and the influence from highlight facula posed on the extraction is obviously suppressed. After that, the relative pose and position determination algorithm is developed based on nonlinear terms transformation. Compared with the nonlinear propagation method, the proposed method can work stably with low computation load. Finally the ground-based monocular vision system is established, and he vision determination experiments are carried out when target is floating slowly. The accuracy and real time performance of the vision determination method are illustrated by simulation results.
出处
《北京理工大学学报》
EI
CAS
CSCD
北大核心
2013年第10期1015-1020,共6页
Transactions of Beijing Institute of Technology
基金
国家自然科学基金资助项目(1102018)
关键词
共面光标
单目视觉
光斑干扰
coplanar feature points
monocular vision
facula interference