Fast and non-destructive discriminating the geographical origin of Hangbaiju by hyperspectral imaging combined with chemometrics

Abstract

Hangbaiju is highly appreciated flower tea for its health benefits, and its quality and price are affected by geographical origin. Fast and accurate identification of the geographical origin of Hangbaiju is very significant for producers, consumers and market regulators. In this work, hyperspectral imaging combined with chemometrics, was used, for the first time, to explore and implement the geographical origin classification of Hangbaiju. The hyperspectral images in the spectral range of 410-2500 nm for 75 samples of five different origins were collected. As a versatile chemometrics tool, bagging classification tree-radial basis function (BAGCT-RBFN), compared with classification tree (CT), radial basis function network (RBFN), was applied to discriminate Hangbaiju samples from different origins. The results showed that BAGCT-RBFN based on optimal wavelengths yielded superior classification performances to CT and RBFN with full wavelengths. The recognition rates (RR) of the training and prediction sets by BAGCT-RBFN were 96.0 % and 92.0 %, respectively. Hyperspectral imaging combined with chemometric can be considered as a powerful, feasible and convenient tool for the classification of Hangbaiju samples from different origins. It promises to be a potential way for origin discriminant analysis and quality monitor in food fields.

Keywords: Chemometrics; Geographical origin; Hangbaiju; Hyperspectral imaging; Quality control.