一种基于OPTICS聚类的仅测角导航目标检测算法

仅测角自主导航方法具有设备简单、复杂度低，功耗低的优点，在空间任务中具有广泛的应用前景.针对中远距离下空间目标特征少的特点，提出了一种利用基于OPTICS聚类算法的空间目标检测方法，可用于仅测角导航过程中的目标检测.对原始星图进行预处理提取星点及目标点，并结合星图识别的结果选择部分帧，使用经过改进的OPTICS聚类方法获得目标运动轨迹.最后，使用本文中的算法对软件仿真出的含有目标的高精度星图进行处理验证了算法的可行性.在卫星相对于空间目标抵近过程中，目标检测的水平误差及垂直误差小于0.15°的帧数分别占到了85.4%以及99.6%.相比AVANTI实验中的目标检测方法，减少了在轨任务中相关参数的调节，进一步提升了算法的自主性.

A Space Target Detection Method Based on OPTICS Clustering for Angles-Only Relative Navigation

The relative navigation using only line of sight (LOS) information has the advantages of simple equipment, low complexity, and low power consumption, and has a wide range of applications in space missions. This paper proposes a space target detection method based on ordering points to identify the clustering structure (OPTICS) clustering in the view of the fact that there are few features of space non cooperative targets during the relative navigation with only angle measurement at medium and long distances. The raw image collected by the star camera is preprocessed to extract star points and target points, and some frames are selected according to the result of star identification. The improved OPTICS clustering method is used to obtain the target trajectory. Finally, to confirm the algorithm's effectiveness, it is used to process the high precision star map containing the target simulated by the software simulation. The proportion of horizontal and vertical errors less than 0.15 degree is 85.4% and 99.6% during the progressive process. Compared with the target detection method in autonomous vision approach navigation and target identification (AVANTI) experiment, the adjustment of related parameters in the on orbit mission is reduced, and the autonomy of the algorithm is improved.