天基监视空间目标轨迹提取算法  

Image processing algorithm for target detection in space-based surveillance

作　　者：奚晓梁[1] 周晓东[1] 张健[1] 

机构地区：[1]海军航空工程学院控制工程系,山东烟台264001

出　　处：《电子设计工程》2011年第19期154-158,共5页Electronic Design Engineering

摘　　要：序列星图中弱小目标的检测与定位是可见光天基监视中的关键技术之一,星图预处理的结果直接影响检测灵敏度及虚警率。文中引入一种在轨检测轨迹提取算法,该算法适用于高斯噪声的星图,由高斯最小二乘拟合、小区域滤波及星象边缘阈值分割三步组成。可抑制噪声背景减少虚假目标的同时,较好地保持星象边缘。通过算法性能分析可知,同美国天基可见关相机(SBV)在轨检测(Moving Target Indicator,MTI)及传统阈值分割算法相比,虚假目标数量减少80%,弱小目标信噪比提高3 dB。由于对星斑边缘保持较好,使得恒星及卫星的定位精度优于MTI算法1个角秒。算法实时性强,有利于工程应用。Detection of weak-small moving objects in time sequence of star images is a problem encountered in space-based surveillance to geosynchronous earth obit. Star image preprocessing is a key technique for getting a good performance of sensitivity and probability of false detection. In this paper, a n algorithm is introduced to target detection, which is composed of three steps： least square matching, small domain filter and threshold segmentation. The algorithm gets a trade-off between sensitivity and background clutter suppression, meanwhile keeps the edges of targets. By algorithm was performance analysis, compared with MTI and normal threshold segmentation algorithm, the number of false detection was reduced by eighty percent, the SNR of small-weak targets have an increase of 3 dB. An estimation of centroid precision shows that the result of adapting the algorithm has a 1 second of arc optimization less than that of MTI algorithm. The algorithm is easily implemented in engineer because of the regularity and real-time of the operation.

关 键 词：星图预处理 小区域滤波 定位精度 背景杂波抑制 

分 类 号：V557.4[航空宇航科学与技术—人机与环境工程]