Video SAR Reconstruction Based on Low-Rank Representation

Video synthetic aperture radar (SAR) has attracted increasing attention in recent years due to its ability to provide continuous images for the scenes of interest. However, practical applications of video SAR are limited by the large amount of data and computational costs involved in imaging. In this paper, we propose a deep unfolding network for reconstructing SAR videos from undersampled echo data. Firstly, we introduce a low-rank representation operator to perform low-rank representation on the video SAR tensor. Then the problem of reconstructing video SAR is modeled as a regularization problem based on low-rank representation and solved by the alternating direction method of multipliers (ADMM) algorithm iteratively. Finally, we unfold the iterative solution into a deep neural network to learn the network parameters and low-rank representation operator from the data. Simulation experiments validate the effectiveness of the proposed method.