Anisotropic image formation based on basic sequential algorithmic scheme and non-quadratic regularization

The emerging synthetic aperture radar (SAR) missions motivate the application of automated processing techniques in obtaining abundant information from the scene for the decision and interpretation due to developments in aeroelectronics. In this paper, we develop a new anisotropic imaging algorithm called BSAS-NQR which is based on basic sequential algorithmic scheme (BSAS) and the non-quadratic regularization (NQR) techniques. Sparse signal representation and canonical scattering center models have been combined to characterize anisotropic scattering. The BSAS technique is used to cluster atoms in order to intelligently reduce the size and mutual coherence of the overcomplete dictionary. Then, the SAR image and the type of canonical scatterers are obtained by NQR simultaneously. Experimental results show the proposed algorithm not only reduces the memory requirements but also retains anisotropic reconstruction performance.