Detection And Tracking Of Small Targets In Clutter

Targets detection and tracking are the most important tasks of surveillance and monitoring radar systems. With the development and wide utilization of stealthy technology, the radar cross section (RCS) of typical military targets, such as aircrafts, missels, tanks, and combat ships, has been greately decreased. The reflected signal from these targets is very weak and usually hidden in strong background clutter and electronic interference, which causes severe challenges to radar detection and tracking. The traditional detection and tracking approaches cannot guarantee reliable detection and tracking performance when encounted with these small targets, as a result, it's of greate importance to investigate new theories and methodologies which can improve detection and tracking performances.In this dissertation, detection and tracking approaches for small targets in clutter are studied, wherein greate efforts are devoted to several key issues, such as"CFAR detector in spiky clutter","TBD procedures", and"optimal waveform design for enhanced detection of small targets". The main contents and contributions are described as following.1) Some detection performance measures including probability of detection, divergence, deflection coefficient, and Kullback-Leibler (KL) divergence are first summarized and exploited to evaluate the detection performances of deterministic signals, random signals as well signals with unknown parameters. Then, the constant false alarm rate (CFAR) and generalized likelihood ratio test (GLRT) detectors for subspace signal detection in compound-Gaussian clutter are studied and an approximate close-form expression for the probability of detection is derived.2) Some modified Kalman filter (KF) and particle filter (PF) based tracking algorithms for nonlinear and non-Gaussian systems are studied. After that, a ship tracking algorithm with airborne radar in glint noise is presented, where the performances of scoring function modified KF and PF are evaluated and compared.3) A goodness-of-fit (GOF) test based CFAR (GOF CFAR) detector is proposed, and exploited to detect extended targets in SAR images. The proposed GOF CFAR detector shows improved performance in non-Gaussian spiky clutter and interfering targets scenarios.4) A efficient dynamic programming based track-before-detect (DP TBD) procedure with reference to a low pulse repetition frequency (PRF) surveillance radar is proposed. The proposed DP TBD procedure can make use of the ambiguous Doppler information to decrease the search areas, and efficiently obain the maximum a posteriori (MAP) estimates of the target trajectory as well as ambiguity number sequence.5) A Rao-Blackwellised PF based TBD (RBPF TBD) procedure for multitarget is proposed. The proposed RBPF TBD method estimates the number of targets and their states simultaneously by estimating the joint multiple probability density (JMPD), and incorporates CUSUM procedure to quickly detect spawning targets, and fixed sample size (FSS) likelihood ratio test to detect disappearing targets in time.6) An approximate close-form expression for the probability of detection of extended targets in waveform-dependent clutter is first derived with regards to the transmitted waveform, which lays a foundation for the subsequent optimal waveform design. Then, methods for designing optimal, suboptimal as well as asymptotic optimal waveforms for detecting rank-one target are presented.In this dissertation, theories and methods for improving the detection and tracking performances of small targets in clutter are studied, wherein major efforts has been devoted to the following aspects:"CFAR detector in spiky clutter","TBD procedures", and"optimal waveform design for enhanced detection of small targets". Several preliminary results have been achieved in this dissertation, and future work will focus on applying the proposed methods to specific radar system to improve the detection and tracking performance of small targets.