Research On Adaptive Detection Methods Of Weak Target In Heavy Clutter Scenario

Adaptive radar target detection technology has become an important research direction in modern radar signal processing,which can effectively suppress the clutter and detect the target.The classical adaptive detection methods usually assume that a moving target stays in a single range cell during the integration time and the clutter is homogeneous Gaussian noise satisfying independent and identical distribution(i.i.d.).In practice,the target may easily cause an across range cell effect when we increase the integration time to improve the radar's ability of detecting weak target.In addition,the clutter is usually heterogeneous and non-Gaussian distributed due to the complexity of the geographical environment and the improvement of radar resolution.Then,the performance of classical adaptive detection methods will drop dramatically.In order to solve the above problems,the dissertation focuses on the adaptive long time coherent integration algorithms for weak target under homogeneous and heterogeneous clutter environment and the adaptive target detection algorithms in compound Gaussian clutter environment.The main research work is summarized as follows:1.In order to detect a moving target with an across range cell effect in heavy homogeneous clutter environment,an adaptive Radon-Fourier transform(ARFT)is proposed.The ARFT can optimally integrate the target's signal with an across range cell effect and suppress clutter simultaneously by introducing adaptive processing to the model matching search,which breakes the restriction that the target stays in a single range cell.Simulation results show that the ARFT can get higher signal-to-clutter-plus-noise ratio(SCNR)than classical long-term coherent integration methods and classical adaptive detection methods,which is of great significance to improve the radar's ability of detecting weak target.2.To address two difficulties for ARFT's real implementation,one is the lack of training samples in a heterogeneous clutter environment,and the other is the high computational complexity in a long coherent time,a sub-aperture adaptive Radon-Fourier transform(SA-ARFT)is proposed.The SA-ARFT divides all coherent pulses into several sub-apertures based on the clutter's degree of freedom to reduce the adaptive processing dimension and perform the parametic model matching search with these sub-apertures,which can significantly reduce the requirements for the i.i.d.training samples and the computational complexity with little performance loss compared with ARFT.Finally,the effectiveness of SA-ARFT is verified by simulation analysis.3.In order to detect a moving target in compound Gaussian clutter environment,a compound Gaussian clutter model with multivariate inverse Gaussian texture(MIG-CG)is firstly established,which can independently describe the spatial and temporal correlation characteristics of clutter.Based on this model,a TMAP-AMF algorithm is proposed to overcome the performance loss of classical adaptive detection methods caused by the clutter power scale difference.To overcome the problem that the performance of TMAP-AMF decreases sharply in the presence of steering vector mismatches,an adaptive matching filter detector based on Maximum posterior probability(MAP)estimation in spatial dimension(SMAP-AMF)is further proposed.Simulation results shows that the TMAP-AMF detector can obtain higher detection ability than typical AMF detectors with the steering vector is completely known,while the SMAP-AMF detector has better detection ability for mismatch signal than other algorithms in the presence of steering vector mismatches.