Synthetic Aperture Radar Sparse Imaging Method For Structured Target

Synthetic aperture radar?SAR?imaging technology has the advantages of all-weather,all-day,long operating distance and so on,which plays an important role in military and civil fields.With the sustained expansion of imaging radar applications,SAR data acqui-sition system gradually develops toward multi-dimensional observation,such as multi-mode,multi-perspective,and multi-polarization.Multi-dimensional observation puts for-ward higher requirements for high-resolution SAR imaging.The existing SAR imaging technology is faced with the problems of sparse aperture observation,target backscatter field description and phase noise interference.Man-made target is one of the important SAR targets.It usually has the characteristics of spatial sparse scattering,clustered struc-ture scattering and object-level parametric scattering.This dissertation studies the high-resolution SAR imaging method for structured target by using its partial scattering prior,combined with precise description of electromagnetic characteristics of target,sparse aper-ture imaging and change detection,phase noise suppression and other key technologies.The main work of this dissertation can be summarized as follows:?1?Polarimetric SAR object-level sparse imaging method for structured targetUnder certain observation conditions,SAR electromagnetic scattering response can clearly reflect the physical properties of structured targets.On the basis of electromag-netic scattering mechanism,the relationship between the object-level parametric scatter-ing model and the backscattering of the structured target is discussed.Then the structured sparse imaging model based on the information of the attribute scattering center?ASC?and the canonical shape feature?CSF?is established.Considering the joint sparsity of polarimetric SAR data in the parametric space of the object-level model,a simultaneous sparse approximation?SSA?based polarimetric object-level imaging method is proposed.This method uses the mixed norm constraint to solve the joint sparse optimization prob-lem by multiple focal underdetermined system solver?M-FOCUSS?algorithm.It can ob-tain the SAR image with accurate physical attribute information.Aiming at the influence of impulse noise on SAR imaging quality in real environment,a robust super-resolution imaging method is proposed based on the polarimetric ASC model.In this method,the?₁norm data fitting model is designed,and the attribute parameters are estimated by the improved complex alternating direction method of multipliers?ADMM?algorithm.The super-resolution processing is realized in object-level.Using the synthetic data and elec-tromagnetic calculation data,it is verified that the proposed algorithm is robust to impulse noise,and can ensure the structural integrity of the target with obvious super-resolution effect.?2?Sparse aperture SAR imaging and change detection method for structured targetWith multiple observations of fixed scene,the sparse aperture SAR imaging and co-herent change detection?CCD?method is studied based on the sparsity of target.Firstly,the sparse aperture SAR signal model under different observations of fixed scene is es-tablished.In order to improve the imaging performance using the prior information of complex structured objects,the sparsity of target is modeled by Gamma-Gaussian hierar-chical Bayesian model,and the clustered structure of target is described by Beta-Bernoulli model.Finally,the approximate posterior of the scattering coefficient vector is learned from the data in the framework of variational Bayesian inference?VBI?.The proposed al-gorithm can retain the weak scatterers and remove the noisy points,so that the subsequent CCD gain can be improved.Considering the problem that traditional CCD performance depends on the imaging quality and the window function of CCD estimator,a Bayesian joint imaging and change detection method is proposed.Firstly,a partial coherence prob-ability model of scattering coefficient and change indicator is constructed,according to the correlation between the scene and change at different observation time.Then,Markov random field?MRF?prior is given to the scene change to maintain the spatial sparsity and local continuity structure.Finally high-resolution image and high-performance CCD are obtained simultaneously under Bayesian inference.The performance of the devised algorithm is verified by simulation data.?3?Structure-aware SC-SAR imaging and autofocus methodBased on the spotlight circular synthetic aperture radar?SC-SAR?system,the structure-aware imaging and autofocus method is studied.Firstly,the principle of SC-SAR system is introduced.Under the SC-SAR signal model,an improved range Doppler algorithm?RDA?is derived.The effectiveness of the SC-SAR system and corresponding imaging algorithm is verified by simulation data.For the defocusing caused by atmospheric distur-bance in practical application,the imaging and autofocus is regarded as a joint optimiza-tion problem,and a sparse imaging algorithm with phase noise suppression is proposed under the variational expectation maximization criterion.The devised method makes full use of the prior of phase noise and structured target.Specifically,the multivariable Von Mises?MVM?model is adopted to enhance the correlation of phase noise along the pulse dimension,and the pattern-coupled prior and MRF prior are used to characterize the en-ergy concentration property and sparsity of the scene.It can achieve high-precision phase error estimation and well focused imaging results.Experimental results of synthetic scene and electromagnetic calculation data of man-made target verify the performance of the proposed algorithm.