Radar Target Detection And Imaging Based On Bayesian Compressive Sensing

Radar imaging technology is an important application of radar.In the traditional method of radar imaging,the quality of the imaging results have been limited by Nyquist sampling theorem,bandwidth of signals and Rayleigh limit,and then Compressive Sensing(CS)theory has solved such problem.On the one hand,it breaks through the limit of Nyquist sampling theorem,which makes the reconstruction of signal possible when downsampled.On the other hand it improves the imaging resolution.However,the existing radar imaging methods based on CS have some shortcomings,such as the high computational complexity caused by large dimension of scence,the selection of parameters setting and the imaging quality under the low SNR.Compared to the reconstruction algorithm based on the norm constraint,Bayesian compressive sensing(BCS)reconstructs the signal with better sparsity and does not need to set the regularization parameter.Therefore,it is very significant to research the radar target detection and imaging based on BCS.In this paper,we mainly discuss the methods of radar target detection and imaging based on Multitask Compressive Sensing(MT-BCS).Firstly,the related work about BCS theories is introduced systematically,and then new imaging models are established for Range-Doppler imaging and Inverse Synthetic Aperture Radar(ISAR)imaging,and solved them by MT-BCS algorithm.Finally to verify the proposed methods by comparative experiments.For the Range-Doppler imaging,firstly,we introduce IAA algorithm and the 2D-SL0 algorithm.The iteration speed of IAA algorithm will be very slow when the dimension of scene is large.Although the 2D-SL0 algorithm can solve this problem,it does not fully exploit the sparsity of scene,and the imaging results is not good enough when the SNR is low.In this paper,by introducing the sparse prior information of distance dimension and azimuth dimension,we establish a new model of imaging,and then adopt MT-BCS algorithm to solve it.Finally,the comparative experiments results show that the proposed method is not only of high resolution,but also has a good performance when the pulse number is less and SNR is low.For ISAR imaging methods,there are some problems similar to that of distance-Doppler imaging.The traditional RD imaging method has low resolution.The short aperture ISAR imaging method based on CS improves the imaging efficiency,but the resolution of another dimension is still limited by the traditional imaging method.Refer to Range-Doppler imaging process,using the sparsity of distance dimension and azimuth dimension by establishing a separable model.Then an ISAR imaging method based on MT-BCS is proposed under MT-BCS framework.Finally,a set of experiments are designed to demonstrate that the proposed method has a high resolution in the case of low SNR and down-sampling.