A modified CoSaMP algorithm for electromagnetic imaging of two dimensional domains

Abstract

The compressive sampling matching pursuit (CoSaMP) algorithm is used for solving the electromagnetic inverse scattering problem on two-dimensional sparse domains. Since the scattering matrix, which is computed by sampling the Green function, does not satisfy the restricted isometry property, a damping parameter is added to the diagonal entries of the matrix to make the CoSaMP work. The damping factor can be selected based on the level of noise in the measurements. Numerical experiments, which demonstrate the accuracy and applicability of the proposed algorithm, are presented.