Synthetic Aperture Radar Imaging Meets Deep Unfolded Learning : A comprehensive review

Synthetic aperture radar (SAR) can obtain highresolution images without being affected by environmental visibility. The compressed sensing (CS) technique is considered to be a strong candidate for simplifying SAR system complexity and improving imaging quality. With CS, SAR imaging is addressed by optimizing an augmented object function with data fidelity and feature-oriented priors, however, suffering from the time-consuming calculation and poor adaptability. Recently, an emerging technique dubbed deep unfolded/unrolled learning, or model-driven learning, offers promise in eliminating such issues by bridging the gap between learning framework and iterative algorithms. The increasing popularity of unfolded networks in SAR inverse problems also show their potential in developing efficient and accurate imaging algorithms. This paper surveys the SAR imaging algorithms based on deep unfolding techniques. We extensively cover different imaging regimes including conventional 2-D SAR, inverse SAR (ISAR), three-dimensional SAR (3-D SAR), and automotive radar imaging. On the algorithm side, the deep unfolding frameworks are mainly categorized according to the feature-oriented regularizers, wherein, their characteristics, principle, and feasibility in SAR inverse problems are discussed in detail. By reviewing the pioneer works, we discuss and reveal the current research stage in different tasks. Finally, the limitations, challenges, and opportunities of deep unfolding techniques are discussed in different radar imaging tasks. © 2024 IEEE.