基于近红外光谱的山核桃品种鉴别

Varieties Identification of Pecan(Carya cathayensis Sarg.)Using Near Infrared Spectroscopy

为实现对山核桃品种的快速鉴别,采集浙江临安山核桃、安徽宁国山核桃、美国山核桃和四川核桃共4种100个核桃样品的近红外光谱,对光谱数据进行了标准正态变量变换(standard normal variate transformation,SNV)预处理后,采用主成分分析(principal component analysis,PCA)法实现了光谱的差异可视化,基本可实现4种山核桃的鉴别。为提高模型准确率,采用主成分分析降维后数据结合线性判别(PCA-LDA)的分类方法,该法对4种山核桃品种鉴别的校正集和验证集的分类准确度都达到了100%。结果表明,近红外光谱技术可实现对山核桃品种的快速鉴别。

To realize varieties identification of pecan(Carya cathayensis Sarg.),near infrared spectroscopy of 100 samples from four kinds of pecans,including Zhejiang Lin'an pecan,Anhui Ningguo pecan,American pecan and Sichuan pecan,were collected.To achieve visualization of spectral differences,principal component analysis(PCA)was carried out on experimental data.It was found that spectral principal component analysis after standard normal variate transformation(SNV)processing could basically identify four different kinds of pecans.To improve accuracy of the model,linear discriminant analysis after principal component analysis(PCA-LDA)was applied to near infrared spectroscopy.It was found that varieties identification accuracy of correction set and verification set reached 100%.Results showed that near infrared spectroscopy could effectively identify pecan varieties.