Generalized MED blind deconvolution of GPR data and its sparsity-promoted solution

Over the past decades, numerous efforts have been attempted to enhance the resolution and accuracy of surface ground-penetrating radar (GPR) data by employing blind deconvolution techniques. The fact that most GPR source wavelets utilized in practice are non-minimum phase presents some challenges for blind deconvolution of GPR data. This letter formulates blind deconvolution of GPR data as a sparsity promoted optimization problem, which extends classical minimum entropy deconvolution (MED) strategy and provides a general-purpose framework of blind deconvolution of GPR data. And then an alternating iterative method is explored to solve the derived non-convex optimization problem. Experimental results demonstrate that the proposed methodology is very effective, efficient and flexible.