基于TSVD的块稀疏重构雷达前视超分辨成像

Radar forward-looking super resolution imaging based on block sparse reconstruction with TSVD

针对传统贝叶斯解卷积方法未能有效利用目标的块结构信息而导致对块目标成像效果不理想的问题,本文提出将模式耦合稀疏贝叶斯学习方法应用于前视成像中。同时,针对其在低信噪比下分辨率迅速恶化的问题,提出使用截断奇异值分解作为解卷积之前的预处理,通过剔除卷积矩阵中较小的奇异值,可以有效抑制解卷积过程中的噪声放大,进而提升方位分辨率。仿真实验结果表明,所提方法在低信噪比下仍具有较好的超分辨成像效果。

In response to the problem of traditional bayesian deconvolution methods failing to effectively utilize the block structure information of the area targets,and resulting in poor imaging performance for area targets.we apply the pattern-coupled sparse Bayesian learning(PCSBL)method to forward-looking imaging in this paper.Moreover,in response to the problem of rapid degradation of resolution in this method at low signal to noise ratio(SNR),we propose to use the truncated singular value decomposition(TSVD)method as the preprocessing before deconvolution,which can effectively suppress the noise amplification in the deconvolution process by discarding the small singular values of convolution matrix.Simulation results show that the proposed method still has good super-resolution imaging effect under low SNR.