Feature-enhanced recovery of low resolution radar imagery based on metrics structured experiment design regularization

We address a new structured descriptive experiment design regularization (DEDR) based approach for feature-enhanced recovery of scene power reflectivity maps from the low resolution imagery acquired with conventional low cost side looking radar systems operating in harsh sensing environments. The feature-enhanced RS scene recovery from the low resolution radar imagery is treated as a nonlinear inverse problem with model uncertainties. To alleviate the problem ill-posedness, we propose to aggregate the error ℓ₂ squared norm minimization based DEDR estimation framework with the variational analysis (VA) inspired - ℓ₂ -ℓ₁ metrics structured (MS) regularization. The MS regularization level is aimed at exploiting the prior model information about the piecewise sparseness of the scene reflectivity gradient maps via aggregating the corresponding metrics structures in the image/solution space. The resulting fused DEDR-MS feature-enhanced sensing method is implemented in the implicit contractive mapping iterative mode. Our DEDR-MS method outperforms the most prominent competing radar image enhancement techniques that do not aggregate the DEDR with the VA inspired MS regularization both in the resolution-enhancement-over-noise-suppression and the convergence rates as verified in the reported computer simulations.