Discriminant analysis of Chinese patent medicines based on near-infrared spectroscopy and principal component discriminant transformation.

Principal component discriminant transformation was applied for discrimination of different Chinese patent medicines based on near-infrared (NIR) spectroscopy. In the method, an optimal set of orthogonal discriminant vectors, which highlight the differences between the NIR spectra of different classes, is designed by maximizing Fisher's discriminant function. Therefore, a model for discriminating a class and the others can be obtained with the tiny differences between the NIR spectra of different classes. Furthermore, because NIR spectra contain a large amount of redundant information, principal component analysis (PCA) is employed to reduce the dimension. On the other hand, continuous wavelet transform (CWT) is taken as the pretreatment method to remove the variant background. For identifying the method, different medicines and the same medicine from different manufactures were studied. The results show that all the models can provide 100% discrimination.