Research on Labeling Method for Fungal Diseases in Murals Based on Hyperspectral Images

Murals are an important part of human cultural heritage, reflecting the cultural characteristics of different periods and containing extremely high research value. However, for a long time, due to environmental factors such as humidity and temperature, as well as internal factors such as mural production materials and processes, mold has grown on murals, seriously affecting their aesthetics and preservation status. Mold disease labeling can detect mold growth and provide technical support for mold prevention and control. Traditional manual annotation methods are not only time-consuming and laborious, but also suffer from subjective errors and inconsistent annotation. Therefore, this article proposes a method for labeling fungal diseases in murals based on hyperspectral images, which can achieve automatic recognition and labeling of fungi on the surface of murals. This method first generates mold training samples based on hyperspectral image unmixing, and then proposes a mural mold labeling network framework based on a multi grouping strategy long short-term memory network and a multi-scale convolutional neural network (MGL-MC), effectively combining spatial and spectral information of mural hyperspectral images. The experiment shows that the method proposed in this article has a good effect on the recognition and labeling of mold on the surface of murals, and can improve the accuracy of mold detection.