Research On Three Dimensional SAR Imaging Based On The Regularization Technique With Penalty

Three-dimensional(3D) linear array synthetic aperture radar(LASAR) imaging obtain along track and cross track resolution through the linear motion of line array antenna, synthesising equivalent virtual two-dimensional array antenna. It obtain high resolution in the direction of the radar line of sight through pulse compression technology. This technique solved the problems in the traditional two-dimensional SAR, which including the shade, top and bottom inverted and facing slope reduction etc.. LASAR array obtain poor resolution along the array because of the limitations of the size of linear array antenna.In the three-dimensional scene, the target is sparse, and can be used for non-related measurement of the compressed sensing technology. LASAR can improve the resolution along the direction of the array by the sparse reconstruction method. The main work and innovation points of this paper are as follows:1. Introduced the theory model of LASAR imaging and compressed sensing, including the linear frequency modulation signal, pulse compression, matching filter, geometric model of LASAR imaging and 3D back projection(BP) algorithm. In this thesis, the theory of compressed sensing is introduced from three aspects: the sparse representation of signal, the construction of sensing matrix and the design of sparse reconstruction algorithm. The orthogonal matching pursuit(OMP) algorithm and basis pursuit(BP) algorithm are introduced in detail.2. Analysed the iteratively reweighted least squares algorithm(IRLS), and putting forward a new sparse reconstruction method: Double threshold sigmoid penalty(DTHS) sparse reconstruction method, and we have DTHS-1 DTHS-2 according the choice of norm. The convergence of DTHS-1 and DTHS-2 algorithm are analyzed in detail, and the convergence process is divided into two stages. In the initial stage of the iteration, the algorithm DTHS-1 degraded into the IRLS algorithm, algorithm DTHS-2 degraded into Tikhonov regularization algorithm. When the iterative sequence of non zero element is greater than the door limit, iterative enters to the dths stage, which ensure the unbiasedness. The parameters used in the DTHS algorithm are analyzed and simulated, and the results show that the upper threshold affecting the performance of the algorithm most. Compared with IRLS algorithm, the DTHS algorithm has better performance. Compared with the OMP algorithm, it is not only suitable for the sparse case, but also for the continuous case.3. Established the LASAR sparse resolution enhancement model. Analysed the sparsity of the scene of LASAR, and constructed three-dimensional linear measurement model according to the radar echo signal equation. Introduced two kinds of modeling method to reduce the dimension of measurement matrix: sparse model based on extracting the strong echo signal and sparse model based on fractal dimension. The simulation experiment obtained the resolutuin enhance ability of the DTHS algorithm. Compared with the traditional back projection(BP) imaging algorithm, the DTHS algorithm has resolution enhancing and sidelobe suppression effect. The digital elevation model(DEM) reconstruction simulation show that the DTHS algorithm is better than the IRLS algorithm, and the IRLS algorithm and DTHS algorithm are better than the BP algorithm.In conclusion, this thesis puts forward the DTHS regularized sparse reconstruction method, and applied to the sparse resolution enhancement of LASAR imaging, provided technical reference for the LASAR image enhancement.