Discrimination of pear varieties using three classification methods based on near-infrared spectroscopy

作者：Fu X; Zhou Y; Ying Y^*; Lu H; Xu H

来源：Transactions of the ASABE, 2007, 50(4): 1355-1361.

• discriminant analysis
• fruit
• near-infrared
• neural network
• partial least squares
• variety classification

Fourier transform near-infrared spectrometry (FT-NIR) was explored as a tool to discriminate samples of three different pear varieties (Xueqing, 'Cuiguan'. and 'Xizilv'), 'which were mainly planted in Zhejiang province, China. Discriminant models were developed using discriminant partial least squares (DPLS) regression, discriminant analysis (DA), and probabilistic neural networks (PNN). Both DPLS and DA calibration models developed in the region of 800-2500 nm gave high accuracy of classification for the calibration set, with 100% of the samples being correctly classified. For validation based on the DPLS model, all 'Cuiguan' pears were classified correctly, whereas a 97.5% classification accuracy was achieved for both 'Xueqing' and 'Xizilv' pears. The DA model had 100% classification accuracy for each variety in validation. For the PNN model, only one sample was misclassified, with an accuracy of 99.2%. The results of this study indicated that FT-NIR had great potential for non-destructively discriminating different varieties of fruits.

• 出版日期2007-8
• 单位浙江大学