Non-line-of-sight 3D-SAR High-precision Imaging Technology Study

The perception of occluded environment targets has significant research significance and application value in fields such as anti-terrorism and riot control,urban warfare,disaster relief,and autonomous driving.Currently,there is an urgent need for low-cost,highresolution,multi-dimensional radar imaging methods that can operate in all-weather and all time.Three-dimensional synthetic aperture radar(3D-SAR)achieves high-resolution threedimensional imaging of targets through antenna arrays and aperture synthesis,making it highly promising for imaging and perception of urban environment targets.However,traditional 3D-SAR based on line-of-sight imaging mechanisms treats scene multipath echoes as interference,making it unable to image non-line-of-sight occluded targets and greatly weakening the imaging and perception capabilities of urban targets.Therefore,breakthroughs in new imaging theories and methods are needed to solve the current mechanism defects of 3D-SAR imaging and achieve high-resolution imaging of non-lineof-sight blind spots in urban environments.Thesis focuses on the high-precision imaging technology of non-line-of-sight 3D-SAR based on electromagnetic multipath scattering,non-line-of-sight imaging,and compressed sensing sparse reconstruction theories.The main research contents of thesis are as follows:1.The basic theory of high-precision imaging of non-line-of-sight 3D-SAR based on multipath utilization is studied.The non-line-of-sight 3D-SAR imaging geometry model is constructed,and the multipath propagation characteristics of non-line-of-sight signals are analyzed.The non-line-of-sight 3D-SAR echo model based on multipath scattering is established,and the echo characteristics of ideal and non-ideal reflection surfaces are analyzed.The observation aperture and imaging resolution properties under non-line-ofsight constraints are explored.The principle of the time-domain back-projection imaging algorithm for traditional line-of-sight 3D-SAR is introduced,and the non-line-of-sight virtual image recognition and extraction method based on morphological detection is studied.2.The sparse imaging method of non-line-of-sight 3D-SAR under ideal reflection surfaces is studied.The sparse characteristics of non-line-of-sight scene targets are analyzed,and a non-line-of-sight three-dimensional imaging model based on compressed sensing reconstruction is established.A mirror sparse three-dimensional imaging(MSSTV)algorithm based on joint L1 norm and total variation(TV)regularization operators is proposed for the sparse characteristics of targets.To solve the problem of large computational complexity of this method,the distance migration reconstruction(RMA)kernel function is introduced,and a mirror sparse total variation imaging algorithm based on RM kernel function(RM-MSSTV)is studied,thus achieving efficient and high-precision imaging of non-line-of-sight 3D-SAR.A ground millimeter-wave 3D-SAR imaging test system is built,and the effectiveness of the algorithm is verified through measured data.3.The self-focusing imaging method of non-line-of-sight 3D-SAR under non-ideal reflection surfaces is studied.The electromagnetic propagation characteristics of non-ideal reflection surfaces and the probability model of echoes are analyzed,and a phase error estimation model of non-line-of-sight 3D-SAR under non-ideal reflection surfaces is constructed.A joint-constraint mirror sparse total variation self-focusing imaging algorithm(MS-RMMSSTV)based on image optimality criteria and RM-MSSTV sparse imaging algorithm is studied to improve the imaging performance of non-line-of-sight 3D-SAR under non-ideal reflection surfaces.The impact of different reflection surfaces on non-lineof-sight radar imaging results is analyzed through measured data,and the effectiveness of the relevant algorithms is verified.