Research On High Resolution ISAR Sparse Imaging Of Complex Moving Targets

The target observed by Inverse Synthetic Aperture Radar(ISAR)is usually the non-cooperative target,and the motion of the observed target relative to the radar in the coherent cumulative time is generally more complex,which greatly increases the difficulty of ISAR imaging and makes the traditional imaging model and algorithm no longer applicable.In this dissertation,the influence of complex moving targets on ISAR imaging is analyzed,and a series of key problems in the sparse imaging process of complex moving targets are solved by using sparse method.The main research work of this dissertation includes:Firstly,the structure sparse imaging method of uniformly rotating target is studied.In order to achieve high resolution ISAR imaging with fewer observational data,a sparse imaging algorithm based on Pitman-Yor process is proposed in this dissertation.The target of ISAR observation often has the sparse characteristics of blocky structure in the image.In order to introduce sparse structure prior into inversion,the image to be inversed is first divided into overlapping grid blocks,and the ISAR imaging problem is transformed into the inversion problem of overlapping grid blocks.The amplitude of grid blocks is endowed with complex Gaussian priors to promote sparsity,and the internal structure information is learned by estimating the covariance of grid blocks.In order to avoid the overfitting problem in the estimation of covariance,Pitman-Yor priors are given to covariance for structural clustering.Due to the clustering power law of Pitman-Yor process,structural blocks with special structures can be retained in the clustering process,making the imaging results more precise.Secondly,the imaging problem of targets with spin components is studied.The scattered echo from the spin component of the target will introduce micro-Doppler signal into the whole echo of the target,which seriously interferes with the ISAR imaging of the main part of the target.In order to suppress the micro-Doppler interference,an inversion algorithm of the range image of the spin component based on dual domain joint sparse is proposed.In the case of Migration Though Resolution Cell(MTRC),the signal from the scattering point of the target body presents a straight line shape in the range image and time-frequency domains,with joint sparse characteristics.The energy of the micro Doppler signal diverges relatively in the range image and time-frequency domain.Therefore,in this dissertation,l2,1 norm is introduced into the inversion problem of signal range images to carry out joint sparse constraints on range images and time-frequency domains.Joint sparse inversion of main signals is constructed.The Alternating Direction Multiplier Method is used.ADMM is used to solve the inversion problem iteratively.When the target has MTRC,this dissertation proves through theory and experiment that micro-Doppler signal will jump to more distance units after correcting MTRC,causing more serious pollution to the image.Therefore,it is necessary to suppress micro-Doppler interference before correcting MTRC.In order to alleviate the weakening of the joint sparsity of the main signal caused by MTRC,this dissertation designed a reweighting method combining different distance units with the same frequency slice to preserve the main signal energy of diverging MTRC in the inversion.Thirdly,the imaging problem of non-uniform rotating target is studied.If the movement of the target relative to the radar can be regarded as non-uniform turntable movement,non-uniform rotation will cause defocus of the image.In this dissertation,the target motion model is phased,the low-order motion model is retained for parameter estimation in the inversion,the high-order motion is approximated as the disturbance term of the observation matrix,and the influence of disturbance is tolerated in the image inversion by sparse population least square method.If the target rotates at a threedimensional non-uniform velocity relative to the radar,the imaging period needs to be optimized due to the unstable imaging plane.In this dissertation,the linear structure inversion algorithm of the main scattering point of the range image is constructed by using the structural continuity and directional continuity of the target main scattering point in the range image,so as to highlight the linear structure.The sub-aperture of the range image is divided and the main scattering lines in each segment are detected by the linear detection method.The main scattering tracks are formed by data association using the Hungarian algorithm.The time-frequency analysis of the unit where the main scattering point is located can finally complete the optimization of the imaging period.Finally,the sparse aperture imaging of complex moving targets is studied.When azimuth sparse sampling is carried out in the echoes of a complex moving target,the absence of Range-Instantaneous Doppler(RID)will cause artifacts and strath interference in the imaging result.Replacing the traditional time-frequency distribution in RID algorithm with sparse time-frequency distribution can solve the artifact problem in imaging.However,most sparse time-frequency distribution is limited by short-time Fourier transform framework,and the time-frequency resolution is low,so the imaging resolution is not high.To solve these problems,a sparse bilinear time-frequency distribution based RID imaging algorithm is proposed in this dissertation.Based on the relationship between the instantaneous autocorrelation function of the sparsely sampled signal and the integral transformation of its Wegener distribution,the inversion problem is constructed,and the sparsity and joint sparsity of bilinear time-frequency distribution are introduced to eliminate the artifact interference.In this dissertation,the iterative inversion of the proposed sparse bilinear time-frequency distribution is given by means of ADMM.