An Accurate Sparse SAR Imaging Method for Joint Feature Enhancement Based on Nonconvex-Nonlocal Total Variation Regularization

Abstract:
Synthetic aperture radar (SAR) imaging under the sparse constraint is a developing SAR imaging scheme that emerged in the recent decade. In sparse SAR imaging, the sparsity can be regarded as a priori information, used to enhance different types of imaging features, among which l1-norm and the total variation (TV)-norm are two widely used constraints in the reconstruction model corresponding to the features of point targets and distributed targets respectively. However, l1 regularization often generates a biased estimation and the TV regularization always over-smooths isolated point targets in the scene. In this paper, we propose the nonconvex-nonlocal total variation (NLTV) regularization, which can realize the joint feature enhancement of sparse SAR imaging. The modified variable splitting-alternating direction method of multipliers (VS-ADMM) is introduced to solve the compound regularization. The performance of the proposed method is verified using simulated data and Gaofen-3 (GF-3) SAR data.