摘要
为提高空间目标天基实时监视能力,对美国在空间中段实验卫星上搭载的空间目标在轨检测与跟踪信号处理器采用的基于时序多帧投影策略的MTI(Moving Target Indicator)算法进行了深入研究,并提出了一种基于时序差分标准化最大值投影的云杂波背景抑制算法。系统地阐述了时序多帧投影算法的思想与典型投影算子的设计方法;针对云杂波成像场景,对MTI算子提出了两点改进方法,即引入相邻帧差分算子与次序统计量中值估计算子来提高对云杂波成像背景的抑制能力,增强背景抑制算法的稳健性;最后,利用实际获取的云层背景图像数据库,仿真比较了不同投影算法的性能差异及影响算法性能的主要因素。仿真实验结果表明:对于输入信杂比小于2.0的弱小目标图像,本文算法的输出信杂比达到5.8以上,比MTI算法平均提高了1.0左右,有效解决了云杂波成像背景的弱小目标检测问题。
To improve the capability of space-based surveillance for space targets, this paper researches the Moving Target Indicator（MTI） based on time-domain projecting image sequences for suppressing starry sky background.The MTI has been used in a spaceborn signal processor on the USA Midcourse Space Experiment for target detection and tracking.According to the in-depth study of MTI algorithm,a new method is proposed to suppress the background of cluttered cloud based on the standardized maximum projection of temporal difference.The mathematical principle and the operator design method of temporal projection are given.Then,considering the statistical characteristics of cluttered cloud,the two Improved MTIs（IMTIs）,namely,the temporal difference operator on an adjacent frame and the estimated median operator,are introduced to the MTI to enhance the clutter suppression.Finally,the experimental method based on the cloud background data acquired is designed to analyze the algorithm performance of different projection methods as well as the main factors affecting on the algorithm.The experimental results show that the output Signal-to-clutter Ratio（SCR） of IMTI reaches more than 5.8 when the input SCR of a small target is less 2.0,which has been improved by 1 in average as compared with that of MTI.It concludes that the IMTI can effectively solve the dim target detection problem of cluttered cloud image background.
出处
《光学精密工程》
EI
CAS
CSCD
北大核心
2012年第4期826-834,共9页
Optics and Precision Engineering
基金
国家863高技术研究发展计划资助项目(No.2006AA704215)
关键词
云杂波背景抑制
时序多帧投影
图像序列
运动目标
cluttered cloud background suppression
time series multi-frame projection
image sequence
moving target
