A Multifrequency Inverse Algorithm for 3-D Microwave Imaging

A new multifrequency inverse algorithm is presented based on the improved variational Born iterative method (VBIM) for 3-D quantitative microwave imaging (MWI). The proposed method is capable of reconstructing the 3-D dielectric relaxation model using measured data acquired at multiple transmit frequencies. A least-squares optimization using the conjugate gradient method with Tikhonov regularization is applied at each VBIM step to relax the memory requirements. For evaluating the algorithm performance, two experiments are conducted with our newly designed MWI system, and image reconstruction is performed under black-box conditions with no a priori information. The phantom models in these experiments are: 1) a single ping-pong ball filled with 75% isopropyl–25% water mixture submerged into a plastic cup filled with 90% isopropyl–10% water mixture and 2) a coplanar distributed object consisting of five ping-pong balls filled with water at the same vertical level. Reconstructed 3-D dielectric images demonstrate the accuracy of our proposed VBIM-based multifrequency inverse algorithm using measured data with no a priori information.