Diagnosing building moisture conditions using CNN analysis of infrared and microwave signatures

Water is a significant cause of cultural heritage deterioration, making moisture detection crucial for heritage conservation. This study utilizes infrared thermography (IRT) in combination with microwave technology to quantitatively and qualitatively assess moisture distribution and dampness-related deterioration in China's Yungang Grottoes. The method accurately diagnoses moisture content and distribution by using a Convolutional Neural Network (CNN) for multi-feature image fusion. Non-destructive techniques and statistical analyses, such as signal-to-noise ratio (SNR) and spatial stability, improve the detection of moisture damage. Statistical analysis confirmed the accuracy of IRT and microwave technology for qualitative and quantitative moisture assessment, providing valuable support for conservation and management strategies.