CLASSIFICATION METHOD OF JAPONICA RICE GEOGRAPHICAL ORIGINS IN HEILONGJIANG BASED ON RAMAN SPECTROSCOPY
作者:Tian, FM (Tian Fang-Ming)[ 1,2 ] ; Yu, HY (Yu Hai-Ye)[ 1 ] ; Tan, F (Tan Feng)[ 2 ] ; Zhao, XY (Zhao Xiao-Yu)[ 2 ]
OXIDATION COMMUNICATIONS
卷: 39  期: 4  页: 3273-3283  子辑: 2
出版年: 2016
查看期刊信息
摘要
A rapid classification method for Japonica rice in Heilongjiang area was established based on the Raman spectroscopy combined with the principal component analysis (PCA) and support vector machine method (SVM) in this paper. There are great differences in the component of Japonica rice due to different varieties and geographical origins. Therefore, it is quite important for the Japonica rice production and trade to build a quick, accurate and effective classification method. 235 samples of Raman spectral lines ranged from 200 to 3400 per centimeter were collected with the Raman spectrometer from the Japonica rice produced in 3 origins of Heilongjiang area. After baseline correction and smooth processing for original Raman data, the Euclidean distance was used to remove the abnormal samples. 20 corresponding values of characteristic peaks were selected by the functional groups analysis for Japonica rice spectrum as the feature vectors. The 2D score chart of the first two principal components was obtained through PCA for 3 types spectrum data of Japonica rice in MATLAB, and a good clustering effect on the 3 different kinds of Japonica rice was shown. For a better classification, a further processing of normalisation needs to be done on the samples. The 2D scores chart for the first two principal components was made based on the normalised data through PCA. The 3 samples were divided into 3 zones and a better clustering effect than that of the former. The original spectrum data were replaced by the score vectors of the first 3 principal components. A C-SVC model based on radial basis kernel function (SVM RBF) was set up after the 100 samples of the three kinds of Japonica rice being trained and the unknown 103 samples being identified. It was shown that the whole accuracy of classification of SVM RBF kernel function for three kinds of Japonica rice is 92.23%, and a good effect was shown by using PCA with SVM method of Raman spectroscopy for Japonica rice classification and identification of different origins in Heilongjiang area.
关键词
作者关键词:Raman spectroscopy; Japonica rice; principal component analysis (PCA); classifying; support vector machine (SVM); geographical origin
KeyWords Plus:MASS-SPECTROMETRY; HUSKS ASH; COMPOSITES; IDENTIFICATION
作者信息
通讯作者地址: Yu, HY (通讯作者)
显示增强组织信息的名称        Jilin Univ, Sch Biol & Agr Engn, Key Lab Bion Engn, Minist Educ, Changchun 130022, Peoples R China.
地址:
显示增强组织信息的名称        [ 1 ] Jilin Univ, Sch Biol & Agr Engn, Key Lab Bion Engn, Minist Educ, Changchun 130022, Peoples R China
显示增强组织信息的名称        [ 2 ] Heilongjiang Bayi Agr Univ, Coll Informat Technol, Daqing 163319, Peoples R China
电子邮件地址:haiyi2009a@163.com
基金资助致谢
基金资助机构        授权号
National High Technology Research and Developmental Program '863'        
2013AA103005-04
National Science and Technology Support Program        
2014BAD06B01
Heilongjiang Province Natural Science Foundation        
F201329
QC2015071
查看基金资助信息   
出版商
SCIBULCOM LTD, PO BOX 249, 1113 SOFIA, BULGARIA
类别 / 分类
研究方向:Chemistry
Web of Science 类别:Chemistry, Multidisciplinary
文献信息
文献类型:Article
语种:English
入藏号: WOS:000392409200004
ISSN: 0209-4541
期刊信息
Impact Factor (影响因子): Journal Citation Reports®
其他信息
IDS 号: EI3RO
Web of Science 核心合集中的 "引用的参考文献": 17
Web of Science 核心合集中的 "被引频次": 0，

已经检索了，是吗？非常感谢！