A Novel ISAR Imaging Method Using Pattern-Coupled Laplacian Scale Mixture Prior

Abstract

We develop a novel method for high-resolution inverse synthetic aperture radar (ISAR) imaging by exploring the block-sparse structure inherent in the ISAR image. First, the Laplacian scale mixture (LSM) prior is utilized to encode the sparsity of the ISAR image. Then, a 2-D pattern-coupled scale prior for sparse coefficients that encourage dependencies among neighboring coefficients is utilized to capture the block-sparse structure of the target scene. Finally, the Laplace approximation-based variational Bayesian inference (VBI) is employed to infer the posterior of the ISAR image along with hyperparameters. Numerical results based on simulated and measured data are provided to validate the effectiveness of the proposed algorithm.