针对近地小天体监测和预警中极低信噪比运动目标难以检测的问题, 提出了基于涨落分析的暗弱运动目标关联检测方法, 阐述了在时间区间上基于光场涨落分析的关联特征提取方法的原理。强背景信号的涨落在时间维度上具有稳定分布和稳定变化的特点, 利用由目标运动导致的瞬态涨落在时间维度上分布的异常变化反演出暗弱运动目标的存在。首先将目标检测问题建模为基于单个像元灰度时序序列的一维信号检测问题, 然后在时域上划分时间窗口, 采用基于涨落分析的方法来提取关联特征, 最后由关联特征的变化来检测目标信号。将该关联检测方法应用到一维时序信号仿真实验、暗弱运动目标仿真实验和近地小行星检测实验中, 结果表明本文所提出方法的检测率和虚警率都显著优于其它几种经典常用方法, 在仿真数据和真实数据中都有很好的检测效果。

To detect moving targets with very low SNRs for monitoring small objects near the ground and providing early warnings regarding such objects, a dim moving target detection method based on fluctuation analysis was proposed. Furthermore, the principle of a correlation feature extraction method based on the fluctuation analysis of a light field in a time interval was expounded. Strong background signals were distributed and changed steadily in the time dimension. Transient fluctuation in the time dimension, caused by the movement of a target, reflected the existence of a weak moving target. First, a one-dimensional signal model based on single pixel gray-scale time series was established. Then, the time window was divided in the time domain, and fluctuation analysis was used to extract related features. Finally, the target signal was detected based on changes in the correlation feature. The method was applied to one-dimensional time-series signal simulation, dim moving target simulation, and a near-Earth asteroid detection experiment. The experimental results show that the detection rate and false alarm rate of the proposed method are significantly better than those of several other commonly used methods. This method provides good detection using simulated and real data.