Improvement of Defect Feature Extraction in Eddy Current Pulsed Thermography

Defect inspection based on eddy current pulsed thermography (ECPT) technique is a research focus in non-destructive testing (NDT) area. In this paper, a novel defect feature extraction approach for automated defect analyzing based on ECPT has been proposed to improve the feature quality of defects. This approach can separate defect features from background interference such as noises and heating coil. The highlights of this algorithm are mainly on the following three aspects: 1) the physical meaning of initial data is combined with the mathematical algorithm to improve the interpretability of results; 2) local sparse component evaluation based on principal components analysis can reserve more significant defect features; and 3) a two-phased fusion criterion for feature integration is defined to improve the quality of defect features. Two typical steel specimens are utilized to testify the validity of the proposed approach. Compared with other three feature extraction algorithms in ECPT, the proposed approach can reserve more useful defect features and suppress more background interferences.