An Effective Signal Processing Procedure for the Detection of Road Damages using Ground Penetrating Radar

Pavement damage is actually one of the most crucial problems in roads. Agencies need first to localize the damage, second to identify the causes. Indeed, the rehabilitation can be compromised, if the cause is not removed. The ground penetrating radar (GPR) technique is used by many agencies involved in roads management. In fact, it is nondestructive and is promising for soil characteristics interpretation, such as moisture or density. A preliminary detection and a subsequent classification of the pavement damage, based on an automatic GPR analysis, have been performed and experimentally validated. An optimum detection procedure is performed. It implements the classical Neyman-Pearson radar test, which is based on a constant false alarm rate strategy. All the setting needed by the optimum procedure have been estimated from a training set of measures. The overall performance has been evaluated by looking at the usual receiver’s operating characteristic (ROC), obtained by processing a different (from the training phase) set of traces. The obtained results evidence that a reasonable performance has been achieved by a suited analysis of GPR images exploiting the spatio-temporal correlation properties of the received signal. Although a generalization is not reliable, this study shows that an automatic GPR-based evaluation of subgrade characteristics is feasible. Key-Words: Detection, Geophysical Signal Processing, Ground Penetrating Radar, Image Analysis, Spatiotemporal signal processing.