Discriminant Analysis of Red Wines from Different Aging Ways by Information Fusion of NIR and MIR Spectra

Abstract

Two different levels’ information fusion of near infrared (NIR) spectra and mid infrared (MIR) spectra coming from red wines aged in different ways has been investigated. A total of 96 red wines, including 44 wines aged in oak barrel, 26 wines aged in stainless steel tank with oak chips and 26 wines aged in stainless steel tank, were analyzed. The NIR transmission spectra and MIR attenuated total reflectance (ATR) spectra of samples were collected. For data level fusion, the NIR and MIR spectra were max-min normalized respectively and then merged. For feature level fusion, five PCs and db3 coefficients of 3-layer wavelet decomposition were calculated. Discriminant models of the three different aging wines were established using DPLS and Fisher method before and after fusion. All the fusion methods mentioned above have achieved better results on classification than individual spectroscopy. These results suggest that the spectral fusion of NIR and MIR is a promising technology for discriminating different aging wines.