Identification and quantification of adulterated Tieguanyin based on the fluorescence hyperspectral image technique

The potential of fluorescence hyperspectral imaging technology (FHSI) (400–1000 nm) for qualitative and quantitative analysis of Tieguanyin (Tie) adulteration was proposed. In this study, various preprocessing methods such as multiplicative scatter correction (MSC), first derivative (1stDer), and second derivative (2ndDer) and their combinations were used for improving the spectral quality. Principal component analysis (PCA) was used for sample data exploration and feature dimensioning. Various machine learning models were used for modeling. The results showed that 1stDer+MSC+Random forest (RF) made an accurate prediction of the type of adulteration of Tieguanyin. For quantitative prediction, both RF and partial least squares regression (PLSR) provided accurate predictions of adulteration levels, resulting in Rp2 values ranging from 0.9804 to 0.9831. The results suggest that FHSI combined with the machine learning method can be used as an effective method to detect tea adulteration. This study provides new methods and ideas for other tea adulteration, and is of great significance for protecting the legitimate rights and interests of consumers and maintaining the order of the tea market.