Regularized resolution enhancement of point-based features of synthetic aperture radar image using variable quasi-norm

Nonquadratic regularization with l_p quasi-norm has been widely used as an efficient and powerful tool for resolution enhancement of point-based features of synthetic aperture radar (SAR) images. However, adjustment of the l_p quasi-norm usually requires a lot of time and labor. In this paper, we propose a modified model and method for choosing regularization term. Considering the sparseness of scatterers in the scene of a SAR image, we use the generalized Gaussian distributions (GGD) as prior distributions for sampled scattering field. Our regularization model is constructed based on variable l_p quasi-norms, and the selection of l_p quasi-norm is achieved through estimation of the shape parameter p of the GGD by adopting a moment method. The regularization model leads to an alternating iterative algorithm. Experimental results with simulated and real data show that the method can automatically select regularization term and produce SAR images with improved spatial resolution.