A Multi-Resolution Evolutionary Programming Technique For GPR Applications

An efficient inverse scattering multi-step technique is proposed for dealing with large-scale high-contrast buried scenarios. This method is based on evolutionary programming (EP) stochastic optimizer that ensure a reliable inversion in high-contrast problems. Moreover, an iterative multi-resolution approach (IMRA) is integrated with EP, to guarantee the convergence of large-scale problems. This algorithm starts with using large meshes in lower frequencies, and then, step by step, tighten the borders of search domain to the estimated target location to increase the resolution, without increasing the number of unknown parameters. This method is assessed by implementing it to a large-scale high-contrast GPR scenario. The presented results show that this technique can be considered as an efficient and reliable inversion technique for such applications. Moreover, comparing IMRA-EP results with IMRA-PSO, indicates that the proposed method significantly outperforms the latter.