Towards Automation of Subsurface Defects Detection in Composites Through Analysis of Thermal Images Using a Hybrid Approach

Active thermal imaging has been attracting a lot of interest as a promising non-destructive technique since it allows the detection of subsurface defects in composites, as well as in other materials. Thermal images captured using this inspection method enable the visualization and localization of structural defects existing within a material. However, determining whether a feature captured in thermal images corresponds to structural defect is not always straightforward, it depends on an inspector’s experience. There is the need to automate the interpretation of thermal image datasets to enhance detection, identification’s accuracy and expedite the inspection of components in several industries making used of thermal imaging methods. Herein, it is described the development of an automated subsurface defect detection system which makes use of a hybrid approach based on two models: a neural network based deep learning model and a signal processing method making use of thermal transport characteristics in solids. This complementary approach aims to reduce the identification of false positives as well as to provide a reliable identification of structural defects based on the thermal properties of composites. The implementation of the defect detection system into a portable infrared imaging device is also discussed.