基于星图识别的空间目标快速天文定位

Fast celestial positioning for space objects based on star identification

本文提出了一种基于星图识别的快速定位算法用于加快空间目标的定位速度和提高定位精度。首先,采集当前图像的拍摄时刻和编码器角度值,运用天文定位三角形建立的地平坐标系与地惯坐标系的转换关系,计算设备视轴在地惯坐标系中的指向;然后,由视轴指向所在的天区,提取出该天区所有的导航星与特征库;结合有初始指向的局部星图识别建立图像中背景恒星星像与导航星的对应关系。最后,根据小孔成像原理,通过已识别的背景参考恒星以角距离匹配的方式对空间目标进行相对定位。对4°×4°视场、分辨率为1 024×1 024的实际图像的试验结果表明,因定位过程中引入了天文定位三角形模型和星图识别,大大加快了空间目标的定位速度,初始定位速度平均约为400ms;又因采用相对定位的方式排除了影响定位精度的因素,空间目标的定位精度优于2″。

A fast celestial positioning method based on star identification was proposed to improve positioning accuracy and reduce positioning time.Firstly,the photographic time and angular values were acquired,and the pointing direction of boresight in geocentric inertial coordinate system was calculated by transforming a horizontal coordinate system into a geocentric inertial coordinate system with an astronomical triangle model.Then,all navigation stars and patterns in the area that the boresight pointed were extracted,and the corresponding relationship between stars in images and in the sky was established by a region star match.Finally,the celestial positioning for a space object was completed relatively by reference stars in a background according to pinhole imaging model.The experiment for an image with a FOV of 4°× 4°and a resolution of 1024 × 1024 shows that the course of the positioning for the space object has been greatly accelerated due to the introduction of astronomical triangle model and region star match,and the positioning speed is approximately 400 ms.On the other hand,the positing accuracy of a space target is better than 2″ because the relative positioning method eliminates a lot of factors effecting the accuracy.