Research On Adaptive Detection Methods Of Weak Moving Targets In Sea Clutter

The main focus of this dissertation is on the research of weak moving target detection against high-resolution sea clutter. As an important branch of radar target detection and the most complex section of radar signal processing, target detection against sea clutter has a far-research effect on the military and civil field. When radar operate in high-resolution mode, the non-stationary and non-Gaussian nature of sea clutter is very obvious, which seriously prevent radar from detecting weak target in sea clutter. Therefore, how to detect these moving targets from high-resolution sea clutter becomes a hot issue in radar community. This dissertation is aimed at the problem arising from the research work of sea clutter and corresponding detection scheme is put forward. The main contents of the dissertation are summarized as follows:The high-resolution sea clutter sequence contains a number of the abnormal cell, which can influence the performance of the traditional adaptive matched filter (AMF) detector. Firstly, the abnormal cell can lift the threshold of adaptive matched filter detector and the weak target cannot be detected; Secondly, the fact that the reference cells include the abnormal cell can cause the singularity of the sample covariance matrix, which makes the detection result trustless. Without lessening the number of reference cells, the power of clutter and the structure of the clutter covariance matrix are estimated by the median of the power of the reference cells and the normalized reference cells, respectively. The multiplication of the above two terms is the so-called the power median and normalized sample covariance matrix (PMNSCM). When the PMNSCM is used in the detection scheme of the AMF, it can be proved that the corresponding detector is asymptotically constant false alarm rate (CFAR). Also, it reduces the effect of the abnormal cell and improves the performance of the traditional AMF detector.The shape-parameter-dependent matched filter detectors are proposed for moving target detection in K-distributed clutter, which are specified by a single parameter α∈[0,1], the α-MF detectors for short. The α-MF detectors include the matched filter (MF) detector and normalized matched filter (NMF) detector as special examples with α=0 and a=1. The parameter a can be chosen to match clutter characteristics. An empirical formula is given that the optimal parameter α equals the number of integrated pulses divided by it plus the shape parameter of K-distributed clutter. It is proved that the a-MF detectors are CFAR with respect to the scale parameter, clutter covariance matrix, and Doppler steering vector. The properties of adaptive a-MF (a-AMF) detectors are discussed. For K-distributed clutter, the optimal a-MF detectors are superior to the MF and NMF detectors and are comparable with the optimal K-distributed detectors (OKDs) of high computational cost. Finally, real high-resolution sea clutter data are available to verify the proposed detectors. The optimal a-AMF detectors under K-distributed clutter model are competitive in performance with the adaptive generalized likelihood ratio test linear threshold detectors (GLRT-LTDs) that are optimal for the compound-Gaussian clutter with the inverse Gamma texture.The subband detection scheme, a frequency division tactic to decompose received time series into low-rate subband time series, and the GLRT-LTDs are combined to form a subband adaptive detector to find weak moving targets in sea clutter by lengthened integration time. To alleviate the conflict between the large number of integrated pulses and limited reference cells constrained by spatial inhomogeneity of sea clutter, a discrete Fourier transform modulated filter bank is used to decompose high-rate sea clutter into low-rate subband clutters. Subband clutters exhibit diversity of non-Gaussianity and subbands are grouped into noise-dominated subbands of approximate Gaussianity, clutter-noise-mixed subbands of non-Gaussianity and clutter-dominated subbands of strong non-Gaussianity. The subband compound Gaussian (CG) model with the inverse Gamma texture is presented to characterize subband clutters, and a bi-percentile method is given to estimate the shape and scale parameters of subband amplitude distributions. The GLRT-LTDs specified by the subband parameters are imposed on individual subbands to optimize detection performance. The experiments show that the subband adaptive GLRT-LTDs attain better performance than the subband adaptive normalized matched filter (ANMF) detector, owing to the full exploitation of the subband diversity of the non-Gaussianity of sea clutter.Long integration is often required to detect weak moving target in sea clutter. However, the Doppler frequency spread and amplitude fluctuation in long integration and limited reference cells resulting from spatial non-homogeneity of sea clutter make the traditional adaptive detector work badly. By observing that the compound Gaussian distribution (CGD) with inverse Gamma texture gives a good fit to sea clutter and the instantaneous frequency is slowly varying, a combined adaptive detector is proposed in the thesis, namely the combined adaptive generalized likelihood ratio test linear threshold detector (CA-GLRT-LTD), which consists of the product of the maximal response of the adaptive GLRT-LTDs in several continuous short integration intervals. Owing to the optimality of the GLRT-LTDs for CG clutter with inverse Gamma texture, the proposed detector obtains better performance than the combined adaptive normalized matched filter (CANMF) detector.