Discrimination of tea varieties using FTIR spectroscopy and allied Gustafson-Kessel clustering

Abstract

For the purpose of classifying tea varieties, allied Gustafson-Kessel (AGK) clustering was proposed to cluster the Fourier transform infrared reflectance (FTIR) spectra of tea samples. As a fuzzy clustering algorithm, AGK can not only produce fuzzy membership and typicality values but also cluster various shapes of data with the help of Gustafson-Kessel (GK) clustering. After FTIR spectra were collected by FTIR-7600 infrared spectrometer, they were preprocessed with multiple scatter correction (MSC). To reduce the dimensionality of FTIR spectra and make the classification of data easily, principal component analysis (PCA) and linear discriminant analysis (LDA) were used to process the FTIR spectra. After that, fuzzy c-means (FCM) clustering, possibilistic c-means (PCM) clustering, AGK clustering and allied fuzzy c-means (AFCM) clustering were performed to cluster data, respectively. The clustering accuracy of AGK achieved 93.9% which was the highest one than other fuzzy clustering algorithms. The results obtained in experiments showed that AGK coupled with FTIR spectroscopy could provide an effective discrimination model for classification of tea varieties successfully.