Reconstruction and characterisation of non-uniform pulsed laser detection of CFRP defects based on time-reversed thermal wave imaging (TRTWI) with multi-image fusion

To address the challenges posed by uneven excitation from a non-uniform large-area pulsed laser and constraints such as the offset angle of thermal camera acquisition, this study proposes a defect reconstruction method based on time-reversed thermal wave imaging (TRTWI). By reversing the time-series temperature data to the initial thermal wave state, the method restores the thermal characteristics of defects under weak excitation and mitigates the interference caused by the background thermal wave under strong excitation. This effectively enhances the original detection performance of non-uniform laser thermal imaging. Building on the reconstructed thermal wave images, a multi-image fusion Fuzzy C-Means (FCM) clustering image segmentation method is introduced. The triple categorisation-based classification method effectively differentiates between the defective region, the heat diffusion region, and the background. Unlike single-image defect edge characterisation, the multi-image fusion approach extracts defect features across different scales, enabling more precise defect segmentation.