High-Resolution 2-D ISAR Imaging Based on Torque-Clustering Fast Interior-Point Method

High-resolution and computationally efficient reconstruction has long been a central theme in sparse inverse synthetic aperture radar (ISAR) imaging. This letter proposes a new high-resolution 2-D ISAR imaging approach based on the torque-clustering fast interior-point method. We first recast the ISAR echo model using a sparsity-enhanced 2-D decoupled atomic norm formulation and obtain a nonconvex, gridless imaging model under the 2-D decoupled atomic norm minimization (2D DANM) framework. A reweighting strategy is then employed to iteratively solve the 2D DANM problem, strengthening sparsity promotion and improving resolution. In each iteration, a fast interior-point method is used to reduce computational burden. After optimization, a 2-D Capon spectrum is constructed to extract all candidate scattering-center pairs. Finally, torque clustering is applied to reliably pair these candidates and identify the true dominant scattering-center pairs. Numerical results demonstrate that the proposed method achieves high-resolution ISAR images while offering markedly improved efficiency for large-scale data.