Line laser infrared thermography detection and characterisation of delamination defects in CFRP based on quasi-static reconstruction and thermal diffusion compensation

To address defect recognition errors caused by image spatio-temporal misalignment and thermal diffusion in line laser scanning infrared thermography, this paper proposes a method integrating spatial correction and thermal diffusion compensation. The method combines a quasi-static reconstruction model based on spatio-temporal alignment with a temperature-difference-driven thermal diffusion compensation mechanism, which significantly improves imaging quality and geometric fidelity of defects. In the reconstruction stage, the SAPR (Spatial-Alignment-based Pseudo-Reconstruction) method corrects the scan-induced spatio-temporal misalignment, with an SNR enhancement of 2.27 dB and a thermal contrast enhancement of about 30.0%. During feature extraction, PCA compresses the temporal matrix. Compared to t-SNE, PCA improves feature sparsity by over 20.0%, reduces computation time by 1.2 s, and decreases defect localisation error by approximately 3 pixels. Based on the compressed features, K-Means clustering is used to realise automatic segmentation of defect regions; to correct the influence of thermal diffusion on defect size estimation, a temperature-diffusion-time mapping model is constructed and calibrated under different laser powers, and the average radius estimation error is reduced from 16.4% to 7.0%, with the maximum error within 20.0%. Validation on multi-depth CFRP delamination confirms high accuracy and stability, enables efficient post-processing inspection and dynamic monitoring of composite structures.