Robust Compressive Terahertz Coded Aperture Imaging Using Deep Priors

Abstract

Terahertz coded aperture imaging (TCAI) is a promising radar imaging technology that leverages coded aperture antenna to achieve forward-looking imaging without relying on relative motion. For solving the target scattering coefficient, one typical approach is to impose generic sparsity prior constraint in the image domain. Although the method has made significant progress in TCAI, it is still challenging to achieve the high-resolution reconstruction of targets with different sparseness under compression measurements, and it is also limited by poor noise resistance performance. To tackle these problems, we propose a new imaging scheme using deep prior captured by the generator. In this letter, we first analyze and model the system based on a coherent detection array and propose a deep prior-based model for TCAI by modeling the target as being in the range of the generator. Then, the deep alternate minimization (Deep-AM) algorithm is designed to solve the model by projecting between the target space and the latent variable space in an alternating fashion. Finally, the simulation results demonstrate the robustness and effectiveness of the proposed approach under the compression measurements. Also, the acquisition and coding times of the system are greatly reduced by utilizing array receiving and deep prior information.