Two-dimensional Pattern-coupled Sparse Bayesian Learning with Variable Coupling Coefficients for Microwave Staring Correlated Imaging

Microwave staring correlated imaging (MSCI) is a novel imaging technique based on temporal-spatial stochastic radiation field, which can realize super-resolution imaging. To exploit the continuity of clustering targets, two-dimensional pattern-coupled sparse Bayesian learning algorithm (PCSBL-2D) and adaptive clustered SBL algorithm (ACSBL) have been proposed, introducing a globally shared parameter to describe the relevance between neighboring coefficients. However, the degree of relevance actually varies between targets in 2D scale, which is ignored by the above algorithms. Therefore, a 2D pattern-coupled SBL algorithm with variable coupling coefficients for MSCI is proposed, namely 2D-VPCSBL. Based on hierarchical Bayesian probabilistic model, the variable coupling coefficient is determined by not only the scattering coefficient and the hyperparameter of the corresponding scattering point but also that of four immediate neighboring points in 2D scale. Experiments illustrate that the proposed algorithm outperforms PCSBL-2D and ACSBL under the condition of super resolution.