摘要
利用近红外光谱分析技术对来自3个省份的水蜜桃进行研究,比较主成分分析-线性判别分析(PCA-LDA)、判别偏最小二乘法(DPLS)、支持向量机(SVM)等方法对光谱数据识别的有效性差异。结果表明,SVM的准确率和召回率均高达94.47%,明显优于PCA-LDA和DPLS,更适用于水蜜桃产地溯源。
In this study,honey peaches from three provinces were analyzed by near infrared spectroscopy analysis technique,and the effectiveness of principal component analysis-linear discriminant analysis(PCA-LDA),discriminant partial least squares(DPLS)and support vector machine(SVM)for spectral data recognition was compared.The results showed that the precision and recall rate of SVM were 94.47%.The SVM method was obviously better than PCA-LDA and DPLS,and it was more suitable for traceability of honey peach origin.
作者
孙晓明
陈小龙
余向阳
卞立平
孙爱东
SUN Xiao-ming;CHEN Xiao-long;YU Xiang-yang;BIAN Li-ping;SUN Ai-dong(Institute of Food Safety and Nutrition,Jiangsu Academy of Agricultural Sciences/Jiangsu Key Laboratory for Food Quality and Safety-State Key Laboratory Cultivation Base of Ministry of Science and Technology,Nanjing 210014,China)
出处
《江苏农业学报》
CSCD
北大核心
2020年第2期507-512,共6页
Jiangsu Journal of Agricultural Sciences
基金
国家重点研发计划项目(2017YFC1601000)
江苏省农业科技自主创新基金项目[CX(18)3054]
国家现代农业产业技术体系桃体系项目(CARS-30-5-03).
关键词
水蜜桃
产地溯源
近红外光谱
主成分分析-线性判别分析
判别偏最小二乘
支持向量机
honey peach
geographical origin traceability
near infrared spectroscopy
principal component analysis-linear discriminant analysis
discriminant partial least squares
support vector machine
