Abstract
Data collected in active infrared thermography (AIRT) experiments for non-destructive defect detection in materials are often contaminated by undesired noise and backgrounds. In this study, an AIRT data processing method, which adopts adaptive fixed-rank kriging, is proposed. This approach computes a set of ordered functions that represent data features at the different resolution levels, called multi-resolution spline basis functions. Multiresolution spline functions were extracted from the thin-plate splines and ordered by the degree of smoothness. The only tuning parameter for this method is the resolution level, making this approach extensively applicable. The performance of the proposed method was evaluated by conducting a mosaic sample defect detection. The results showed that the proposed AIRT data processing method is not only efficient but also effective.