高光谱结合支持向量机鉴别不同产地丹参药材  

Identification of Salvia Miltiorrhiza Regions by Hyperspectrum and Support Vector Machine

作　　者：孙成玉 焦龙[1] Sun Cheng-yu;Jiao Long(College of Chemistry and Chemical Engineering,Xi'an Shiyou University,Xi'an,Shaanxi 710065,China)

机构地区：[1]西安石油大学化学化工学院,陕西西安710065

出　　处：《福建分析测试》2023年第2期11-15,共5页Fujian Analysis & Testing

基　　金：西安石油大学研究生创新与实践能力培养项目(批准号:YCS21211036)资助。

摘　　要：采用高光谱结合支持向量机方法(SVM)建立了不同产地丹参药材的鉴别方法。采集了6种不同产地丹参药材的高光谱;之后,分别使用均值中心化和Savitzky-Golay平滑滤波2种光谱预处理方法,结合SVM建立丹参产地鉴别模型;Savitzky-Golay平滑滤波方法结合SVM分类效果最佳,测试集分类准确率为97.50%,同时具有更高的真正率、命中率、和特异度。研究结果表明,建立的高光谱技术结合支持向量机方法步骤简便、准确、可靠,是一种很有前景的丹参药材分析鉴别方法。Hyperspectrum combined with support vector machine(SVM)method was established to identify Salvia miltiorrhiza samples from different geographical regions.In the experiment,the hyperspectra of 6 kinds of Salvia miltiorrhiza samples from different regions were collected.Then a discriminant model was established by SVM method combined with 2 different spectral preprocessing methods(Mean centralization and Savitzky-Golay smooth derivative).The Savitzky-Golay smooth derivative combined with SVM had the best discrimination effect,and the test set classification accuracy was 97.50%.The overall results showed that hyperspectral technique combined with support vector machine method was a promising method for the analysis and identification of Salvia miltiorrhiza.

关 键 词：高光谱 支持向量机 定性分类 中药 丹参 

分 类 号：O657.3[理学—分析化学]