GPR tomography based on regularization method for concrete defect detection

GPR is a nondestructive detection method which has been widely applied to detect defects in concrete structures. In field geotechnical engineering cases, it is very hard to observe anomalies caused by longitudinal defects using ground-based GPR detection, which produces great difficulties in data processing and interpretation. Compared with the traditional ground-based GPR, the cross-hole GPR can obtain more objective image results because of greater exploration depth and higher resolution. Therefore, it is convenient to analyze both defects' distribution and structure of the concrete with cross-hole or hole-ground GPR measurements. In this paper, the minimum support stabilizing functional (MS) is applied to cross-hole GPR travel-time tomography to image defects in concrete structures based on the Tikhonov regularization theory and conjugate gradient inversion algorithm. 2D synthetic data of concrete defects and a field experimental cross-hole data set have been inverted using the MS stabilizing functional. Compared with the minimum norm stabilizing functional (MN) and the maximum smoothness stabilizing functional (maxsm), the new proposal can make the defects' boundary become more obvious, and the inversion results are more close to the real model.