Abstract:

Diffuse reflectance Fourier transform infrared spectroscopy (DR-FTIR) combined with soft independent modeling
of class analogy (SIMCA) was used to study pattern recognition of sticky rice, long-shaped rice and polished round-grained
rice, and the corresponding models were developed by leave-one-out cross-validation based on such pre-treatments as ninepoint
Savitzky-Golay smoothing, baseline correction and standard normal variate (SNV) normalization in the wavelength range
of 1000－1750 cm-1. The results showed that all the SIMCA models were valid, and the identification rates and rejection rates
of the prediction set samples were both 100% under the significance level of α = 0.05 indicating DR-FTIR combined with
SIMCA to be an effective strategy for pattern recognition of rice varieties for Chinese yellow wine making.

Key words: Chinese yellow wine, rice, diffuse reflectance Fourier transform infrared spectroscopy (DR-FTIR), soft independent modeling of class analogy (SIMCA), principal component analysis (PCA)