A GPU accelerated architecture for real-time GPR image reconstruction

The Ultra-Wideband (UWB) Synchronous Impulse Reconstruction (SIRE) Ground Penetrating Radar (GPR) is a vehicle-mounted radar system developed by the US Army Research Laboratory to enable better imaging of landmines and improvised explosive devices. At the Howard University Electromagnetic Imaging Lab, we have proposed novel and parallelizable imaging algorithms that are well-suited to the UWB-SIRE-GPR system. In this paper, we present our work on a GPU accelerated architecture that provides true, real-time GPR image reconstruction from real data collected by the UWB-SIRE-GPR system. Preliminary results show that our solution is economical, powerful, and applicable to a wide variety of other GPR systems.