Enhanced One-Bit SAR Imaging Method Using Two-Level Structured Sparsity to Mitigate Adverse Effects of Sign Flips
Shaodi Ge,Nan Jiang,Dong Feng,Shaoqiu Song,Jian Wang,Jiahua Zhu,Xiaotao Huang
DOI: https://doi.org/10.1109/tgrs.2024.3474712
IF: 8.2
2024-10-25
IEEE Transactions on Geoscience and Remote Sensing
Abstract:One-bit synthetic aperture radar (SAR) imaging technology has garnered significant attention due to its ability to substantially reduce system complexity and data storage requirements. However, noise during the sampling and transmission process can introduce sign flip errors in the echoes, leading to numerous artifacts and clutter in the imaging results, which severely degrade imaging performance. To mitigate this issue, this article proposes an enhanced one-bit sparse imaging method that leverages two-level structured sparsity based on sparse logistic regression (TSS-SLR). The TSS-SLR method employs a soft coupling mechanism grounded in local structured sparsity, linking each element of the sparse signal with its neighboring elements in the original 2-D image space, which enhances the clustered features of the targets in the imaging results. During the iteration process, traditional thresholding steps struggle to effectively suppress high-intensity artifacts and clutter caused by sign flip errors. A threshold selection strategy based solely on pixel level may lead to partial recovery or fragmentation of the target. To address this, the TSS-SLR method integrates global structured sparsity constraints through superpixel (SP) segmentation and weighted hard thresholding operations, further suppressing artifacts and clutter. This technique does not alter the actual amplitude of the elements, allowing weak scatterers in the target region to receive higher weights influenced by strong scatterers within the same SP, thereby enhancing retention and improving imaging quality. For efficient model optimization, the alternating direction method of multipliers (ADMMs) is utilized, enabling separation and alternating optimization of various constraints. Finally, extensive simulations and real measured data experiments validate the effectiveness of the TSS-SLR method for one-bit SAR imaging in the presence of sign flip errors.
imaging science & photographic technology,remote sensing,engineering, electrical & electronic,geochemistry & geophysics