Generative Deep Learning-Based Thermographic Inspection of Artwork

Infrared thermography is a widely utilized nondestructive testing technique in the field of artwork inspection. However, raw thermograms often suffer from problems, such as limited quantity and high background noise, due to limitations inherent acquisition equipment experimental environment. To overcome these challenges, there growing interest developing thermographic data enhancement methods. In this study, defect inspection method for based on principal component analysis proposed, incorporating two distinct deep learning approaches enhancement: spectral normalized generative adversarial network (SNGAN) convolutional autoencoder (CAE). The SNGAN strategy focuses augmenting thermal images, while CAE emphasizes enhancing their quality. Subsequently, (PCT) employed analyze processed improve detectability defects. Comparing results using PCT alone, integration led 1.08% signal-to-noise ratio, utilization resulted an 8.73% improvement.