Identification of fumigated and dyedLycium barbarumby hyperspectral imaging technology

Abstract

In view of the adulteration of Lycium barbarum with chemicals and dyes in the market, hyperspectral imaging technology is proposed to identify fumigated and dyed specimens. The hyperspectral data were preprocessed by visual molecular dynamics (VMD) to reduce the influence of noise. Furthermore, competitive adaptive reweighted sampling (CARS) was used to select the characteristic bands from the preprocessed spectral data. Since the classification effect of the traditional support vector machine (SVM) model was influenced by the kernel parameters c and g, the slime mold algorithm was adopted to optimize SVM. The optimal c and g were set to 93.3603 and 4.5652, respectively. The results showed that the VMD-CARS-SVM had higher prediction accuracy, and the accuracy of training and testing sets were 98.2 and 96.7% separately.