Prediction-correction method for electromagnetic imaging with current reconstruction

A prediction-correction methodology is presented for improving the numerical stability of the iterative process involved in the solution of nonlinear inverse scattering problems related to the electromagnetic imaging of dielectric objects. In various methods developed so far, each iteration is performed by only using the results obtained in the preceding iteration. The results computed in the preceding iteration along with those acquired in previous iterations are employed selectively to make a prediction of the object function to be used in the next iteration. As a consequence, the numerical stability and the convergence of the iterative process are improved. This is illustrated by numerical computation results obtained for lossless and for lossy dielectric bodies