Research On Moving Infrared Small And Dim Target Detection Under Clutter Cloud Image Background

For space-based infrared (IR) ballistic missile defense system (BMDS) ongeosynchronous orbit (GEO), one of the most important techniques is the movingdim target detection under cluttered cloud image background. This paper wouldresearch on the algorithms of cloud background suppression and trace extraction ofmoving target, which are essential contents of moving target detection.In this paper, we developed the projection method and applied it to thecluttered cloud suppression of GEO BMDS. Assuming the platform is stable, andthe velocity of the point target in the focal plane is more than1.0pixels per frame,we developed the optimal temporal difference (OTD) operator based on the optimallog likelihood ratio detector test (LLRDT) method. Then the OTD operator wasimproved to be the standardized maximum projection of temporal difference onadjacent frame (SMPTDAF) operator due to the constraints of small number ofsamples and high real-time of spaceborn systems, exploratory data analysis methodwas also used as an important part during this process. Experimental results showthat the proposed algorithm gives a substantial signal clutter ratio (SCR) gaincompared with the Max-Median filters and the High-Pass filters.If the maneuverability of the moving target is weak, the trace extraction ofwhich could be considered as looking for linear streak in3-D spaces. In order tohaving the capability of space-based surveillance for space target, a spacebornsignal processor of target detection and tracking was applied to the USA MidcourseSpace Experiment (MSX). The MSX moving target indicator performed the traceextraction in a two-stage,"nominator-confirmer" procedure, the first stageperforming a binary matched filter to "nominate" streak paths which looked likethey might actually contain signal and the second performing a energy accumulationdecision to "confirm" the nominations of the first stage. Based on the in-depth studyof above procedure, an improved method was proposed to detect moving targets. In"nominator" stage, velocity restraint condition was introduced to binary matchedfilter for reducing the probability of false alarm. In "confirmer" stage, the criteria of"minimum sample mean difference" was added to that of "maximum sample mean"for improving the probability of detection. Finally, the experimental method was designed to analyze the algorithm performances of two different methods. Theexperimental results show that streak paths number filtered by improved method islower more than a half compared with former method, and moving targets aresevered from candidate paths more easer. Improved method can effectively detectmulti-moving targets really-time.The software for detection and tracking of moving dim targets based onVC++6.0simulation environment was developed in the last section. While targets tobe detected satisfy some conditions on some aspects, such as brightness andorientation, the software could capture them quickly and track loop-locked, whichvalidates the effectiveness of our new detection scheme in some sense.