Inverse Profiling of Inhomogeneous Subsurface Targets With Arbitrary Cross Sections Using Covariance Matrix Adaptation Evolution Strategy

The problem of subsurface inverse profiling of a 2-D inhomogeneous buried dielectric target is addressed in this letter. An iterative optimization technique is proposed that utilizes Covariance Matrix Adaption Evolutionary Strategy (CMA-ES) as its inverse solver and Method of Moments, using Conjugate Gradient-fast Fourier transform, as the forward solver. The numerical results indicate that CMA-ES, as its first reported implementation in buried target reconstruction, can successfully be applied to this challenging reconstruction problem. Also, comparison with Evolutionary Programming and Particle Swarm Optimization indicates that CMA-ES can significantly outperform the other two-optimization techniques in the inhomogeneous subsurface imaging. In addition, examples of various scenarios involving noisy data, lossy targets and multiple targets further demonstrate that CMA-ES can be considered as a robust, simple, and efficient optimization tool in the reconstruction of complex buried targets.