Rapid detection of synthetic pigments in black tea using hyperspectral imaging technology and machine learning

In this study, we propose a novel approach that integrates hyperspectral imaging with machine learning to address the challenge of rapid and non-destructive detection of artificial pigment adulteration in Wuyi Congou tea. Compared with traditional chemical methods, this method enables the identification of pigment types, quantification of adulteration rate, and visualization of pigment distribution. A hyperspectral dataset of black tea adulterated with three synthetic pigments (Tartrazine, Carmine, and Sunset Yellow) was constructed, supporting the establishment of qualitative identification and quantitative prediction models. The qualitative model achieved an accuracy greater than 98 %, while the quantitative model yielded a predictive correlation coefficient (R_p) greater than 0.95 and root mean square error of prediction (RMSEP) less than 0.025. These results validate the feasibility and effectiveness of this approach for detecting and quantifying pigment adulteration in black tea, providing new possibilities for tea quality and safety inspection.