基于近红外高光谱成像结合分水岭算法鉴别酸枣仁药材的产地<sup>&#x0002A;</sup>

张璐, 茹晨雷, 殷文俊, 郑洁, 张慧, 颜继忠

doi:10.16155/j.0254-1793.2021.04.22

药物分析杂志 >

2021 , Vol. 41 >Issue 4: 726 - 734

DOI: https://doi.org/10.16155/j.0254-1793.2021.04.22

快速分析

基于近红外高光谱成像结合分水岭算法鉴别酸枣仁药材的产地^*

展开

1.浙江工业大学药学院,杭州 310014;
2.浙江大学机械工程学院流体动力与机电系统国家重点实验室,杭州 310027

第一作者 Tel：18058794060;E-mail：573134782@qq.com

** 张慧 Tel：（0571）88320984;E-mail：zh889@zjut.edu.cn
颜继忠 Tel：（0571）88320506;E-mail：yjz@zjut.edu.cn

修回日期: 2020-06-23

网络出版日期: 2024-07-15

基金资助

* 浙江省自然科学基金项目（No. Y20H280079）

收起

Identification of Ziziphi Spinosae Semen from different habitats based on near-infrared hyperspectral imaging technology and watershed algorithm^*

Expand

1. College of Pharmaceutical Sciences,Zhejiang University of Technology,Hangzhou 310014,China;
2. State Key Laboratory of Fluid Power and Mechatronic Systems,College of Mechanical Engineering,Zhejiang University,Hangzhou 310027,China

Revised date: 2020-06-23

Online published: 2024-07-15

Fold

摘要

目的：基于近红外高光谱成像结合分水岭算法建立一种快速、无损、绿色的鉴别酸枣仁药材产地的方法。方法：对不同来源的2 215 个酸枣仁样品进行高光谱扫描,从高光谱数据中提取相应的光谱和图像信息,并采用分水岭算法对聚集的酸枣仁样品进行目标分割识别,实现单粒样本平均光谱的自动提取。进一步比较了一阶导数（derivative 1）、二阶导数（derivative 2）、多元散射校正（multiplicative scatter correction,MSC）、Savitzky-Golay 平滑（S-G smoothing）和标准正态变量变换（standard normal variate,SNV）5种不同预处理方法对建模的影响。同时建立了最小二乘法判别分析（partial least squares discrimination analysis,PLS-DA）、支持向量机（support vector classification,SVC）、随机森林（random forest,RF）3种不同的判别模型。并对所建立的模型性能采用准确率、混淆矩阵（confusion matrix）、受试者工作特征曲线（receiver operating characteristic curve,ROC）和曲线下面积（area under the curve,AUC）这4 个指标进行评价。结果：二阶导数是最有效的预处理方法,对预处理后的数据采用PLS-DA 建立的模型较优,其训练集、验证集和测试集的准确率分别为99.87%、99.27% 和99.14%,并且其混淆矩阵、ROC 曲线和AUC 均显示了该模型对于酸枣仁药材产地分类的优越性。结论：本研究建立的近红外高光谱成像技术结合分水岭算法对酸枣仁药材产地的鉴别能力较强,可为工业化在线检测方法的开发提供技术支撑。

关键词： 高光谱成像; 近红外; 分水岭算法; 酸枣仁; 产地鉴别; 最小二乘法判别分析; 支持向量机; 随机森林

本文引用格式

张璐, 茹晨雷, 殷文俊, 郑洁, 张慧, 颜继忠 . 基于近红外高光谱成像结合分水岭算法鉴别酸枣仁药材的产地^*[J]. 药物分析杂志, 2021 , 41(4) : 726 -734 . DOI: 10.16155/j.0254-1793.2021.04.22

Abstract

Objective: To establish a quick,non-destructive and green method for identifying the habitat of Ziziphi Spinosae Semen（ZSS）based on the near-infrared（NIR）hyperspectral imaging and watershed algorithm. Methods: A total of 2 215 ZSS samples from different habitats were scanned by near-infrared （NIR）hyperspectral technology. The spectral and image information of ZSS from the hyperspectral data were extracted. Combining with the watershed algorithm,the aggregated ZSS samples could be segmented and identified to realize the automatic extraction of the average spectrum of a single sample. The effects of five different pre-processing methods on the establishment of discriminant models were compared,including 1st derivative,2nd derivative,multiplicative scatter correction（MSC）,Savitzky-Golay smoothing（S-G smoothing）and standard normal variate（SNV）. In addition,three different discrimination models,partial least squares discrimination analysis（PLS-DA）,support vector classification（SVC）and random forest （RF）,were established and compared. The capability of these models was evaluated by four indicators: accuracy rate,confusion matrix,receiver operating characteristic curve（ROC）and area under the curve （AUC）. Results: 2nd derivative was an effective pre-processing method,and the model established by PLSDA was superior to the pre-processed data. The accuracy rates of the training set,validation set,and test set were 99.87%,99.27%,and 99.14%,respectively. The results showed confusion matrix,ROC curve and AUC exhibited the superiority of the model. Conclusion: In this study,the established method that applied NIR hyperspectral imaging technology combined with watershed algorithm illustrates a strong ability to identify the habitat of ZSS,which can provide technical support for the development of industrial online detection.

Key words： hyperspectral imaging technology; near-infrared; watershed algorithm; Ziziphi Spinosae Semen; habitat identification; partial least squares discrimination analysis; support vector classification; random forest

参考文献

[1] 中华人民共和国药典2015 年版.一部[S].2015:366
ChP 2015.Vol ⅠhP 2015.Vol Ⅰ[S].2015:366
[2] 王玉,杨雪,夏鹏飞,等.酸枣仁汤化学成分、药理作用、临床应用的研究进展及质量标志物的预测分析[J].中国中药杂志,2020,45(12):2765
WANG Y,YANG X,XIA PF,et al.Review of chemical constituents pharmacological effects and clinical applications of Suanzaoren decoction and prediction and analysis of its Q-markers[J].China J Chin Mater Med,2020,45(12):2765
[3] 周赛男,陈安家,郭宝林,等.不同产区酸枣仁中斯皮诺素、6-阿魏酰斯皮诺素、酸枣仁皂苷A、酸枣仁皂苷B的含量测定[J].中草药,2019,50(11):2712
ZHOU SN,CHEN AJ,GUO BL,et al.Determination of two flavonoids and two saponins in Ziziphi Spinosae Semen from different producing areas[J].Chin Tradit Herb Drugs,2019,50(11):2712
[4] 常广璐,李国辉,李天祥.天津产与市售野生酸枣仁的质量比较研究[J].中草药,2015,46(5):751
CHANG GL,LI GH,LI TX.Comparison on quality of wild Ziziphi Spinosae Semen between marketed and produced in Tianjin[J].Chin Tradit Herb Drugs,2015,46(5):751
[5] 杜晨晖,闫艳,朱羽尧,等.山西不同产地酸枣仁中脂肪酸类成分的GC-MS 分析[J].中国实验方剂学杂志,2018,24(24):19
DU CH,YAN Y,ZHU YY,et al.Analysis of fatty acid components in Ziziphi Spinosae Semen from different growing areas in Shanxi province by GC-MS[J].Chin J Exp Tradit Med Form,2018,24(24):19
[6] 白文明,王来兵,成日青,等.近红外高光谱成像技术在药物分析中的研究进展[J].药物分析杂志,2018,38(10):1661
BAI WM,WANG LB,CHENG RQ,et al.Research advance in pharmaceutical analysis based on near-infrared hyperspectral imaging technique[J].Chin J Pharm Anal,2018,38(10):1661
[7] 刘金欣,郭徵,李耿,等.新疆地区药用红花的高光谱特征研究[J].中国中药杂志,2013,38(9):1335
LIU JX,GUO Z,LI G,et al.Hyperspectral characteristics of Carthamus tinctorius in Xinjiang region[J].China J Chin Mater Med,2013,38(9):1335
[8] 黄慧,张德钧,詹舒越,等.干贝水分检测的建模及分级方法[J].光谱学与光谱分析,2019,39(1):185
HUANG H,ZHANG DJ,ZHAN SY,et al.Research on sample division and modeling method of spectrum detection of moisture content in dehydrated scallops[J].Spectrosc Spect Anal,2019,39(1):185
[9] RU CL,LI ZH,TANG RZ.A hyperspectral imaging approach for classifying geographical origins of Rhizoma Atractylodis Macrocephalae using the fusion of spectrum-image in VNIR and SWIR ranges(VNIR-SWIR-FuSI)[J].Sensors,2019,19:2045
[10] TIAN X,FAN SX,HUANG WQ,et al.Detection of early decay on citrus using hyperspectral transmittance imaging technology coupled with principal component analysis and improved watershed segmentation algorithms[J].Postharvest Biol Technol,2020,161:111071
[11] 第五鹏瑶,卞希慧,王姿方,等.光谱预处理方法选择研究[J].光谱学与光谱分析,2019,39(9):2800
DIWU PY,BIAN XH,WANG ZF,et al.Study on the selection of spectral preprocessing methods[J].Spectrosc Spect Anal,2019,39(9):2800
[12] TANKEU S,VERMAAK I,CHEN WY,et al.Differentiation between two“ Fang Ji” herbal medicines,Stephania tetrandra and the nephrotoxic Aristolochia fangchi,using hyperspectral imaging[J].Phytochemistry,2016,122:213
[13] TANKEU S,VERMAAK I,CHEN WY,et al.Hyperspectral imaging and support vector machine:a powerful combination to differentiate black cohosh(Actaea racemosa)from other Cohosh species[J].Planta Med,2018,84（6-07):407
[14] 邵琦,陈云浩,杨淑婷,等.基于随机森林算法的玉米品种高光谱图像鉴别[J].地理与地理信息科学,2019,35(5):34
SHAO Q,CHEN YH,YANG ST,et al.Identification of maize seed varieties based on random forest and hyperspectral technique[J].Geogr Geo Inf Sci,2019,35(5):34
[15] 鲍一丹,陈纳,何勇,等.近红外高光谱成像技术快速鉴别国产咖啡豆品种[J].光学精密工程,2015,23(2):349
BAO YD,CHEN N,HE Y,et al.Rapid identification of coffee bean variety by near infrared hyperspectral imaging technology[J].Opt Precis Eng,2015,23(2):349
[16] 王磊,覃鸿,李静,等.近红外高光谱图像的宁夏枸杞产地鉴别[J].光谱学与光谱分析,2020,40(4):1270
WANG L,QIN H,LI J,et al.Geographical origin identification of Lycium barbarum using near-infrared hyperspectral imaging[J].Spectrosc Spect Anal,2020,40(4):1270
[17] 夏梦琴,周建,王静,等.基于分水岭分割的图像处理算法研究[J].科技视界,2019(17):71
XIA MQ,ZHOU J,WANG J,et al.Research on image processing algorithms based on watershed segmentation[J].Sci Technol Vision,2019(17):71
[18] CHOI J Y,HEO S,BAE S,et al.Discriminating the origin of basil seeds(Ocimum basilicum L.)using hyperspectral imaging analysis[J].LTW-Food Sci Technol,2020,118:108715
[19] 张初,刘飞,章海亮,等.近地高光谱成像技术对黑豆品种无损鉴别[J].光谱学与光谱分析,2014,34(3):746
ZHANG C,LIU F,ZHANG HL,et al.Identification of varieties of black bean using ground based hyperspectral imaging[J].Spectrosc Spect Anal,2014,34(3):746
[20] 吴立周,王晓慧,王志辉,等.基于随机森林法的农作物高光谱遥感识别[J].浙江农林大学学报,2020,37(1):136
WU LZ,WANG XH,WANG ZH,et al.Crops identification based on hyperspectral data and random forest method[J].J Zhejiang A F Univ,2020,37(1):136
[21] 林珑,吴静珠,刘翠玲,等.基于模型集群的东北/ 非东北大米产地高光谱鉴别方法研究[J].光谱学与光谱分析,2020,40(3):905
LIN L,WU JZ,LIU CL,et al.Study on hyperspectral identification method of rice origin in northeast/non-northeast China based on conjunctive model[J].Spectrosc Spect Anal,2020,40(3):905
[22] 吉海彦,任占奇,饶震红.基于高光谱成像技术的不同产地小米判别分析[J].光谱学与光谱分析,2019,39(7):2271
JI HY,REN ZQ,RAO ZH.Discriminant analysis of millet from different origins based on hyperspectral imaging technology[J].Spectrosc Spect Anal,2019,39(7):2271

Options

文章导航

模态框（Modal）标题

摘要

本文引用格式

Abstract

参考文献