单 位：
河南农业大学/国家烟草栽培生理生化研究基地
关键词：
可见-近红外光谱技术;烤烟;品种鉴别;偏最小二乘判别分析
摘 要：
提出了一种使用可见光-近红外光谱技术快速、无损鉴别烤烟品种的新方法。应用可见光-近红外光谱技术,采集获得4个不同烤烟品种的160个样本光谱特征曲线,选取全波段光谱(350~2 500 nm)经Savitzky-Golay平滑滤波、多元散射校正和标准正态变换优化处理,建立了烤烟品种的偏最小二乘判别(PLS-DA)模型。结果表明,该模型校正及验证结果与样本分类变量的决定系数R2大于0.971,校准均方根误差(RMSEC)和预测均方根误差(RMSEP)均低于0.096,模型对验证集中40个未知的烤烟品种样本的正确识别率均达到100%。该方法具有很好的分类和鉴别效果,说明应用可见光-近红外光谱技术结合PLS-DA可以用来对烤烟品种进行鉴别。
译 名：
Identification of Flue-Cured Tobacco Varieties Based on Visible and Near-Infrared Spectroscopy Technique
作 者：
Wang Yiding;Zhao Mingqin;Fu Bo;Yue Heng;Henan Agricultural University,National Tobacco Cultivation and Physiology and Biochemistry Research Center;
关键词：
Visible and near-infrared spectroscopy technique;;Flue-cured tobacco;;Variety identification;;Partial least squares-discriminant analysis(PLS-DA)
摘 要：
A new fast non-destructive method for the identification of flue-cured tobacco varieties was developed by means of visible and near-infrared spectroscopy technique. One hundred and sixty samples of four different kinds of tobacco varieties were measured to obtain characteristic curve of spectra by visible and near-infrared spectroscopy technique. Through Savitzky-Golay smooth filtering,multiplicative scatter correction and standard normal variate optimization process to the original full-wave band spectra( 350 ~ 2 500nm) of each tobacco variety,the partial least squares-discriminant analysis model was established. The results showed that the determination coefficients( R2) of classified variables with calibration and verification set results of PLS-DA model were all more than 0. 971,and the root mean square error of calibration( RMSEC)and root mean square error of prediction( RMSEP) were all lower than 0. 096. The PLS-DA procedure was then applied to classify forty unknown tobacco samples of validation set,and the correct recognition rate reached 100%. This method possessed good classification and identification effects,which indicated that visible and near-infrared spectroscopy technique combined with PLS-DA could be used to identify tobacco varieties.
