Comparison of contrast source inversion based methods for solving two dimensional electromagnetic inverse scattering problems

Abstract

Contrast source inversion method (CSIM) is a well-known and effective microwave imaging method. CSIM has been modified to improve the accuracy and convergence of the method: multiplicative regularized CSIM (MR-CSIM), where the total variation is used for regularization, and cross-correlated CSIM (CC-CSIM), where the cost functional is updated to include the cross correlation of the state and data errors. In this work, CSIM, MR-CSIM, and CC-CSIM are examined and compared in terms of efficiency and accuracy for noiseless and noisy measurement data by applying them to two dimensional inverse scattering problems. It is shown that MR-CSIM performs better on noiseless data and CC-CSIM is more immune to noise than MR-CSIM and CSIM. © 2020 Elsevier B.V., All rights reserved.