Feature extraction and analysis for automatic characterization of impact damage in carbon fiber composites using active thermography

The use of composite materials in the industry has increased in the past few decades due to their high strength and stiffness to weight ratios. However, composite materials have a serious weakness: their sensitivity to impact damage. This work proposes a method to automatically characterize impact damage in carbon fiber composites using active thermography. Shape and amplitude features are extracted from the defects detected using image processing. The analysis of these features provides relevant conclusions about their relation to the impact energy, and the influence of number of plies and the type of core of the composite. Finally, a classifier based on neural networks is proposed to automatically characterize the detected defects caused by impact damage according to impact energy. Tests carried out over several specimens that contain impact damage of different energies show excellent performance for the classifier.