Efficient and Robust Sparse Aperture ISAR Imaging Based on Log-Sum Minimization

• Published in: IEEE sensors journal Vol. 25; no. 13; pp. 25777 - 25789
• Main Authors: Yang, Haozhe, Zhou, Xin, Zhao, Zhiyuan, Wang, Luyao
• Format: Journal Article
• Language: English
• Published: New York IEEE 01.07.2025; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Algorithms; Apertures; Approximation algorithms; Compressive sensing (CS); Image reconstruction; Imaging; Incoherence; Inverse synthetic aperture radar; inverse synthetic aperture radar (ISAR) image; iterative log-sum thresholding (ILT); Matching pursuit algorithms; Minimization; Optimization; Parameter modification; Radar imaging; Random sampling; Real time; Signal processing algorithms; Signal reconstruction; sparse aperture; Sparse matrices; Sparsity; Thresholding (Imaging)
• ISSN: 1530-437X; 1558-1748
• DOI: 10.1109/JSEN.2025.3569224

Inverse synthetic aperture radar (ISAR) imaging of sparse aperture is a challenging problem. Noting that the ISAR image generally exhibits strong sparsity, compressive sensing (CS), and sparse signal recovery methods are widely applied for sparse aperture data. However, the existing sparse-aperture ISAR imaging algorithms are either computationally heavy or require manual adjustment of too many parameters, which limits their applications in the real-time ISAR imaging system. This article proposes a high-precision and computationally efficient ISAR imaging algorithm for a sparse aperture. A generalized CS model with log-sum minimization is first considered for ISAR imaging, and then the iterative log-sum thresholding (ILT) algorithm is utilized to solve the optimization problem, where no large-scale matrix inversion is performed to reduce the computational overhead. Specifically, to efficiently estimate ISAR image sparsity and avoid manually tuning parameters, a modified version of the ILT algorithm is proposed, in which an adjustment strategy of the tradeoff parameter <inline-formula> <tex-math notation="LaTeX">\lambda </tex-math></inline-formula> is adopted. To analyze the influence of different measurement matrices on the imaging algorithm, four sparse sampling patterns, including centralized sampling (CES), block centralized sampling (BCS), equally spaced sampling (ESS), and random sampling (RAS), are analyzed based on the mutual incoherence property (MIP). Experiments based on both simulated and measured data substantiate that the proposed two algorithms can achieve well-focused ISAR images and are very efficient to implement. In particular, the imaging time of the modified algorithm is less than one second for a matrix dimension of size <inline-formula> <tex-math notation="LaTeX">{256} \times {256} </tex-math></inline-formula>, which meets the real-time requirements.