Electromagnetic imaging of 2-d inhomogeneous dielectric objects by an improved spectral inverse technique

Abstract

Recent developments of numerical analysis on inverse scattering problems are reviewed. It begins with brief derivations of both inverse scattering algorithms formulated in the configuration and the spectral domains by applying the moment-method procedures for forward scattering problems in the reverse sequence. Compared with the inversion algorithm in the configuration domain, the separation of physical parameters involved in the spectral inverse scattering formulation helps to find proper regularization methods for the reduction of the noise effect. The Fourier series expansion to the field distribution and averaging process in each enlarged cell provide excellent noise reduction performance. ？ 1990, IEEE. All rights reserved.