Discoloration Evaluation of Stone Cultural Heritage Based on Hyperspectral Imaging

Abstract

ABSTRACT Discoloration poses a serious threat to stone cultural heritage, making accurate assessment crucial for its protection and repair. Current detection methods are limited by coverage, slow speed and direct contact with cultural heritage. To address these issues, a spectral index based on the correlation between spectral reflectance and brightness was developed to identify discoloration boundaries in stone cultural heritage. Color difference ΔE* was then calculated to assess the deterioration of sandstone cultural heritage such as Leshan Giant Buddha over recent years. The results show that a strong correlation between the spectral index and the color changes on the Buddha’s face, with the three-band index using 530, 550, and 555 nm showing the best results in identifying discoloration areas. Among the evaluation models, the Competitive Adaptive Reweighted Sampling-Support Vector Machine Regression (CARS-SVR) model effectively quantifies the degree of discoloration. Brightness (L*) shows the strongest response to black pollution on the Buddha’s face. Discoloration is divided into five levels, with severe discoloration concentrated in the nose area, consistent with field survey findings. This study demonstrates the feasibility of using hyperspectral technology for quantitative discoloration assessment in stone cultural heritage.