Modified Compressive Sensing-based Migration for GPR Probing of Reinforced Concrete Bridge Decks

The ground penetrating radar (GPR) nondestructive evaluation (NDE) method often uses the attenuation map to evaluate the health condition of the reinforced concrete bridge decks. The attenuation map is commonly generated by using traditional migration techniques, which require Nyquist-rate time samples of the received data and fine spatial sampling. Recent progress in the compressive sensing (CS) theory and its development in the application of GPR provide an alternative migration method for high-resolution and less cluttered images with a small amount of CS measurements for sparse point-like targets, which can be satisfied in GPR probing of reinforced concrete bridge decks in the two dimensional sense. However, the previous imaging method for GPR landmine detection using the CS-based migration mainly focuses on the target detection and localization, but fails to correctly preserve the relative reflection amplitudes. As the attenuation mapping is the major network-level evaluation tool for bridge deck evaluation using GPR, the relative intensities of the rebar reflections are of vital importance for this application. A modified CS-based imaging method with segmentation is proposed in this paper to solve this problem. The proposed approach can not only preserve the comparative amplitudes, but can also highly decrease the processing speed and the requirement of the computer memory. The effectiveness of the proposed approach is demonstrated using GPR numerical simulation data in this paper.