Speaker
Description
Active infrared thermography is widely used in non-destructive testing due to its non-contact nature and high efficiency. However, its application to microcrack detection on highly reflective and thermally conductive metal surfaces remains challenging due to weak thermal contrast and background interference. In this paper, a novel feature extraction framework based on thermal signal spatial quasi-static reconstruction is proposed to enhance the detectability of microcracks from infrared image sequences. The method transforms spatiotemporal thermal responses induced by a moving laser source into a temporal sequence equivalent to pulsed thermography, enabling the application of advanced feature extraction techniques. Fourier Transform, Principal Component Transform, and Partial Least Squares Transform are employed to extract defect-sensitive features from the reconstructed data. A three-dimensional heat transfer model is developed to analyze the thermal behavior around microcracks and guide the reconstruction process. Experiments are conducted on a TC4 titanium alloy specimen with cracks ranging from 2 µm to 100 µm using a custom-built mobile laser scanning thermography system. Results demonstrate that the proposed method reliably detects cracks as narrow as 2 µm. Quantitative evaluation using signal-to-noise ratio (SNR) shows that feature extraction algorithms significantly improve defect visibility, particularly for suboptimal cracks. The combination of quasi-static reconstruction and feature extraction offers a promising solution for in-situ and miniaturized infrared inspection of microcracks in metallic components.
