Faster 3D GPR Imaging Using 2D Dictionaries with Applications to Sparse Reconstruction

Ground penetrating radar (GPR) has become an established technology for military appli-cations, geological applications and shallow subsurface characterization. Some applications such as utility de-tection or landmine detection require three-dimensional data acquisition. However, 3D GPR imaging is time re-quired for a full 3D migration which typically requires high-speed parallel CPUs and expensive data process-ing packages. As a consequence, 3D manual GPR data acquisition is timeconsuming, labour-intensive. Also it has a high dictionary size problem. This research demonstrates a new approach for 3D GPR imaging that uses the special feature of 2D target space to generate the 3D image. This work aims at the reduction of memory usage and attenuation of maths operations while generating the 3D GPR image.