Nondestructive damage evaluation of composites based on terahertz and X-ray image fusion

Abstract

Inspection of thin and multi-depth delamination and foreign inclusion defects inside glass fiber reinforced polymer (GFRP) composites is a challenge for traditional non-destructive testing (NDT) techniques. Terahertz (THz) and X-ray computed tomography (CT) are two prevalent NDT methods, which also have their unique advantages and relative weakness individually. Specifically, in our previous study, it is found that THz NDT method has a higher contrast and X-ray CT NDT method has a higher resolution for detecting defects in GFRP composites. In this paper, a new fusion algorithm applied to the THz and X-ray CT NDT imaging data, based on the combination of saliency region analysis (SRA) and wavelet based multi-scale transforms (W-MST) is proposed to detect delamination and inclusion defects of GFRP composites. Moreover, the strategy of weighted least square optimization (WLSO) is adopted to eliminate the effects of unregistered images. An exhaustive number of combinations (36 in total) of the proposed fusion algorithm, which include the selection of fusion rules for saliency map, wavelet decomposition levels and wavelet basis functions, have been quantitively compared using objective evaluation indices such as standard deviation (SD) and spatial frequency (SF) to quantify the improvement of detection effect in terms of contrast and resolution. Among them, there are five optimal combinations applicable to five pairs of different source images, and the averages of the SD and SF indicators of fused images have increased by 126% and 190% compared with the source images, respectively. The inspection efficacy of defects using the proposed fusion method has been analyzed visually, showing that the new approach can effectively accentuate the complementary advantages of THz and X-ray NDT methods, resulting in quantifiably improved defect inspections.