Radar Target Detection Methods In Compound-gaussian Sea Clutter

Maritime radars will inevitably receive echoes from sea surface and various targets.There are many kinds of targets and the characteristic of echoes from sea surface is complicated.Therefore,detecting targets effectively in the complicated sea clutter background has always been a research hotshot.The sea clutter from low-resolution radars or at high grazing angles is usually modelled by the Gaussian model.Nevertheless,the sea clutter from highresolution radars or at low grazing angles no longer obeys the Gaussian model and shows strong non-Gaussian characteristics.In this case,the traditional adaptive detectors for the Gaussian model will have the high probability of false alarm(PFA)or the low probability of detection(PD).Moreover,the reference cells,which are used to estimate the clutter covariance matrix,in nonhomogeneous clutter environment,is not sufficient,the detection performance of the adaptive detectors will be seriously degraded.Therefore,in order to improve the detection performance of radar targets in non-Gaussian sea clutter,the following work and contributions have been done in this paper:1.In view of the heavy-tailing behavior of non-Gaussian sea clutter and the serious performance loss from the insufficient reference cells,the adaptive detection of point-like targets embedded in compound-Gaussian sea clutter with the unknown and persymmetric covariance matrix is studied.In the generalized-Pareto-distributed clutter and compound-Gaussian clutter with inverse Gaussian texture(CG-IG clutter),adaptive detectors are derived based on the two-step generalized likelihood ratio test(GLRT),Rao test and Wald test,respectively.The proposed detectors use the a-priori distribution of clutter texture and the persymmetric structure of the speckle covariance matrix to improve the detection performance.The constant false alarm ratio(CFAR)property and detection performance of the proposed detectors are verified by the theoretical analysis and simulated experiments.The detection performance of the proposed adaptive detectors is better than that of the existing detectors.2.Aiming at the problem that adaptive detectors depend heavily on the number of reference cells in non-Gaussian sea clutter,the adaptive detection for point-like targets is studied based on the basis of a-priori distributions of clutter.The speckle covariance matrix is modeled as a random matrix obeying the inverse complex Wishart distribution and the clutter texture follows the inverse Gaussian distribution.Two adaptive detectors are deduced based on the GLRT and maximum a posteriori(MAP)estimate in the compound Gaussian clutter.The designed detectors combine the a-priori distributions of the texture and the speckle covariance matrix.One of two adaptive detectors does not depend on the data from reference cells,while the other combines a-priori knowledge and information from reference cells.The experimental results show that the proposed two detectors have better detection performance than the existing detector in the case of deficient reference cells.Moreover,the adaptive detector based on the weighted covariance matrix estimator is superior to the existing detector for different number of reference cells.3.Aiming at the problem that the test statistics of the optimal coherent detector in CGIG-distributed sea clutter can not be effectively implemented in radar systems because of the inclusion of the modified Bessel function of second kind,a fast and achievable approximate optimal coherent detector is proposed.The K and CG-IG distributions belong to the exponential decay function,and the detection threshold in radar target detection with the low PFA is usually determined by the tail of clutter distribution when the radar parameters are fixed.Therefore,we extend the near-optimum detector ?-MF in K-distributed clutter to realize the near-optimum detection in CG-IG clutter.By deducing the mathematical relationship between the damped exponentials of K and CG-IG distributions,the formula for calculating the control parameter ? in CG-IG clutter is given.The experimental results show that the detection performance of adaptive ?-MF and adaptive optimal coherent detector in CG-IG clutter is very close,and its performance is better than that of adaptive normalized matched filter(ANMF)and adaptive matched filter(AMF).4.In view of the fact that the existing sea clutter amplitude distributions can not fit the measured non-Gaussian sea clutter in some cases,the compound Gaussian distribution with generalized inverse Gaussian texture(CG-GIG Distribution)is obtained via using the generalized inverse Gaussian distribution to model the clutter texture.Moreover,the optimum coherent detection in the CG-GIG clutter background is studied.The generalized inverse Gaussian distribution contains the Gamma distribution,the inverse Gamma distribution,and the inverse Gaussian distribution,so the CG-GIG distribution can fit the non-Gaussian clutter well.Moreover,the optimum coherent detector is also derived in CG-GIG clutter according to the two-step GLRT.The CFAR property and the ability of the proposed detector is verified.Theoretical analysis and simulated experiments verify the CFAR properties with regard to the speckle covariance matrix and the Doppler steering vector.The simulated and measured data shows the detection performance advantages of the proposed detector.