Electromagnetic imaging of conducting cylinders by applying a genetic algorithm

Abstract

Detection of the shape of perfect conducting objects from information contained in their scattering data is formulated as an inverse problem in terms of nonlinear integral equations. The difficulties in obtaining acceptable reconstructed images lie in the nonlinear nature and in the ill-posedness of the associated inverse problem. Various algorithms have been proposed based on the physical optics approximation, as well as on the exact electromagnetic field equations. To overcome the ill-posedness of this inverse problem, an optimization procedure is usually implemented, where the shape of the conducting object is reconstructed by minimizing the root-mean-square error of the difference between the predicted and the measured data, subject to certain constraints or a priori information. Newton-Kantorovitch method, Levenberg-Marquardt algorithm, and conjugate gradient techniques axe typical deterministic optimization schemes which are used for inverse problems. These are local optimization methods and their efficiency strongly depends on the initial guess.