一种GEO空间目标快速识别算法

An Automatic Identification Method for GEO Space Objects

由于GEO(地球同步轨道)空间目标相对背景恒星的视运动速度较慢,使用相邻帧图像差分的方法难以自动识别并跟踪目标。基于Lucas-Kanade算法,使用在全图所有星像的邻域开窗,计算统计波门内星像的移动速度,根据目标的运动特征给定全局阈值判别的方法,实现了相邻帧短曝光图像间的GEO目标自动识别与跟踪。仿真实验表明,该算法稳健可靠,星像位移计算精度为10-3,计算时间快于0.1s,在观测数据的实时处理中有较大的应用价值。

It＇s difficult to detect and track GEO（Geostationary Earth Orbit） objects automatically using only two short exposure time frames with difference method due to the relatively slow velocity between background stars and GEO space objects.A technique based upon Lucas-Kanade algorithm is presented to resolve this problem.First,specific gates are set around all star images on the frame.Then,the moving velocity of all gates is obtained.Finally,the GEO objects are identified according to the specific movement features using the global statistical threshold.Simulation and tests demonstrate that the method works effectively with robust performance.The displacement precision of object images is about 10-3 and the computing time is less than 0.1s,showing a high value for realtime data processing.