Application of Near Infrared Spectroscopy in Rapid Identification of Celastrus Varieties
近红外光谱技术在快速鉴别南蛇藤品种中的应用

A method for rapid non-destructive identification of Celastrus varieties by near-infrared spectroscopy is proposed. Six kinds of Celastrus are collected. Their spectra in the region from 12493 cm-1 to 4000 cm-1 are obtained with a spectrometer. Through the clustering analysis of pre-processed spectral data by principal component analysis (PCA), ten principal components are obtained. Then, several variety identification models are established by combining the PCA with different stoichiometries. Since the score disribution in PC1 and PC2 has a remarkable clustering effect for different samples, it can be used to discriminate Celastrus varieties qualitatively. 165 samples randomly selected from 220 samples are used as modeling sets so as to establish linear discrimination analysis (LDA), back-propagation artificial neural network (BPANN) and support vector machine (SVM) models. The remaining 55 samples are used to validate the prediction. After optimization of principal components, all models have an identification precision up to 100%. The result shows that the proposed method is effective in the classification and identification of Celastrus varieties.