Study on Equal Interval Sampling Method Promoting Efficiency of Post-Processing Algorithms for Infrared Thermography Inspecting Carbon-Fiber Reinforced Plastics Composites
Qiang Wang,Ruicong Xia,Qiuhan Liu
DOI: https://doi.org/10.1134/s106183092360020x
2023-10-04
Russian Journal of Nondestructive Testing
Abstract:The broadening use of carbon-fiber reinforced plastics (CFRP) in various industries has raised safety concerns. To control the hazard, infrared thermography inspection techniques, capable of large-area scanning, instantaneous imaging, and completely non-contact detection, has been increasingly adopted to detect flaws inside CFRP structures. And in practical engineering, post-processing algorithms are required to eliminate noise and enhance defect detectability after initial inspection course. However, rapid updates of infrared camera bring high-resolution thermograms yet. Whereas classic algorithms have encountered data explosion problem which led to low efficiency and engineering infeasibility. Therefore, to promote the efficiency of postprocessing algorithms, we propose equal interval sampling (EIS) method to sample raw thermal data. Two CFRP laminates with embedded artificial Teflon inserts of different sizes at different depths are inspected by a laser infrared thermography system. The obtained raw thermal videos made up of 500 frame thermal images with 512 × 640 pixels were converted to 3D thermographic data. We apply EIS to sample the data by setting different intervals. And the sampled data are fed to classic post-processing algorithms, thermal signal reconstruction (TSR) and principal component analysis (PCA), which have prevailed in processing thermographic data. Regarding different intervals of EIS as variables, the processing efficiency, and the defect detection performance of EIS-TSR and EIS-PCA are studied. The results demonstrate that EIS-TSR method fails to promote processing speed and maintain the performance of defect detection at the same time. In contrast, the EIS-PCA method attains a remarkable efficiency and intact defect detectability simultaneously, manifesting its adequacy in engineering practice.
materials science, characterization & testing