Coarse-to-fine Estimation: Compressive Sensing for High-Resolution Inverse SAR Imaging

Abstract:
Compressive sensing (CS) enables imaging with less data, assuming sparsity in certain domains. We proposed a groundbreaking coarse-to-fine (CTF) estimation method that reconstructs images with fewer signal supports over a fine grid. Owing to the explicit storage and usage within sparse reconstruction, the sensing matrix (SM) is resource-intensive, even for small problem sizes. To alleviate this issue, the CTF procedure discards a certain number of columns from the SM as the reconstruction progresses from coarser to finer scales. A comprehensive experimental validation was performed using ISAR datasets of both real targets and targets of known geometry, and the qualitative and quantitative metrics of the imaging results confirmed the efficacy of retrieving high-resolution ISAR images. We substituted a single SM with an exorbitant size with a sequence of SMs of reduced size, thereby reducing computational demands and making the technique suitable for real-time applications. The CTF method reconstructed the target’s image that was free from clutter and effectively decoupled the required number of measurements from the ISAR data, thus achieving super-resolution. The entropy of the reconstructed image in the fine stage proves the completeness of the imaging.