A new method for forward-looking scanning radar imaging based on L1/2 regularization

Abstract

The performance of scanning radar forward looking imaging mainly depends on the improvement of azimuth resolution which can be realized by deconvolution. However, due to the ill conditioned nature of deconvolution, it will lead to noise amplification. The regularization method can effectively suppress the ill conditioned of deconvolution by adding regular constraints. Among many regularization methods, L1 constraint is usually used if the target is sparse, but it is not the most sparse solution. Theoretically, L1/2 can produce more sparse solutions than L1. With the development of L1/2 constraint regularization mathematical theory and the improvement of computer computing ability, it can be solved quickly even if it is a nonconvex, nonsmooth and non-Lipschitz optimization problem. In addition, L1/2 constraint regularization has been applied in the field of compressed sensing. In this article, we introduce L1/2 constraint regularization into forward-looking imaging, and the simulation results show that it can achieve better recovery effect than L1 constraint regularization in sparse scene.