Microwave Staring Correlated Imaging of Coupled Block Targets Based on Autoencoder Networks

Microwave staring correlated imaging (MSCI) is a novel imaging technique for super-resolution imaging based on temporal-spatial stochastic radiation field. For the common clustered block sparse targets, traditional sparse Bayesian learning (SBL) based methods including pattern-coupled SBL algorithms (PCSBL) and adaptive clustered SBL algorithms (ACSBL), fail to get the satisfactory result since the coupled block targets can destroys the sparsity of the target. In this paper, an autoencoder network is proposed to achieve information extraction and reconstruction of coupled block targets. The encoder is concerned with extracting the target feature-containing echo information, and the decoder performs the target distribution mapping output. Finally, the proposed method is evaluated using simulated data. Simulation results have demonstrated that under low signal-to-noise ratio (SNR), the proposed method outperforms the traditional sparse Bayesian method significantly for coupled block targets.