PDF下载 +分享

微信公众号：大豆科学

[1]刘瑶,谭克竹,陈月华,等.基于分段主成分分析和高光谱技术的大豆品种识别[J].大豆科学,2016,35(04):672-678.[doi:10.11861/j.issn.1000-9841.2016.04.0672]
　LIU Yao,TAN Ke-zhu,CHEN Yue-hua,et al.Variety Ｒecognition of Soybeans Using Segmented Principal Component Analysisand Hyperspectral Technology[J].Soybean Science,2016,35(04):672-678.[doi:10.11861/j.issn.1000-9841.2016.04.0672]

点击复制

基于分段主成分分析和高光谱技术的大豆品种识别

《大豆科学》[ISSN:1000-9841/CN:23-1227/S] 卷: 第35卷期数: 2016年04期页码: 672-678 栏目: 出版日期: 2016-07-20

Title:: Variety Ｒecognition of Soybeans Using Segmented Principal Component Analysisand Hyperspectral Technology

作者:: 刘瑶¹; 2; 谭克竹¹; 陈月华¹; 王志朋¹; 谢红²; 王立国²; 1．东北农业大学电气与信息学院，黑龙江哈尔滨150030; 2．哈尔滨工程大学信息与通信工程学院，黑龙江哈尔滨150001

Author(s):: LIU Yao¹; 2; TAN Ke-zhu¹; CHEN Yue-hua¹; WANG Zhi-peng¹; XIE Hong²; WANG Li-guo²; 1． College of Electric and Information，Northeast Agricultural University，Harbin 150030，China; 2． College of Information and Communication Engineering，Harbin Engineering University，Harbin 150001，China

关键词:: 高光谱; 分段主成分分析; 大豆; 品种识别

Keywords:: Hyperspectral; Segmented principal component analysis; Soybean; Varity identification

分类号:: TP391. 41; TP274. 5

DOI:: 10.11861/j.issn.1000-9841.2016.04.0672

文献标志码:: A

摘要:: 为了实现大豆品种的快速且无损鉴别，对大豆高光谱图像中的光谱信息进行研究分析。利用高光谱图像采集系统采集波长范围为400 ～ 1 000 nm 的6 类共660 粒大豆样本的高光谱图像，从每粒大豆样本的中心区域上提取感兴趣区域并以此区域的平均光谱信息代表此粒大豆的光谱信息。对光谱曲线进行多元散射校正( multiple scatteringcorrection，MSC) 后，根据相关系数矩阵图，将整个高光谱波段分解为3 个子分段，分别在每个子分段上做主成分分析( principal component analysis，PCA) ，提取1 ～ 20 个主成分作为光谱特征，利用极限学习机( extreme learning machine，ELM) 和随机森林( random forests，ＲF) 模型进行大豆品种识别。结果表明: 在第二分段( 510． 6 ～ 685． 4 nm) 进行PCA变换，识别效果优于全波段PCA 变换。因此，应用分段PCA 变换和高光谱技术对大豆品种进行无损识别是可行的。

Abstract:: In order to realize rapid nondestructive recognition for soybeans varieties， spectral information of hyperspectral imagefor soybeans is analyzed． Hyperspectral images for 660 soybeans including 6 varieties from 400 to 1 000 nm are acquired byhyperspectral image system． The interested region for each sample is extracted and average spectral information is obtained．Spectrum curve is conducted multiple scattering correction ( MSC) ． According to the correlation coefficient matrix，the entirehyperspectral bands are segmented into three highly relevant sub-segments． Principal component analysis ( PCA) is respectivelyused in each sub-segments． The first twenty principal components are extracted as spectral features of soybean samples．Identification model is developed using Extreme Learning Machine ( ELM) and Ｒandom Forests ( ＲF) models． Experimentalresults indicate that the identification results using PCA in the second sub-segment ( 510． 6 － 685． 4 nm) is outperform than usingPCA in the entire bands． Nondestructive recognition and classification for soybean varieties using segmented PCA and hyperspectraltechnology is effective．

参考文献/References:

［1］李文霞，李柏云，薛红，等．黑龙江省不同生态区大豆品种育种性状的主成分分析［J］．大豆科学，2013，32( 6) : 731-734． ( LiW X，Li B Y，Xue H ，et al． Principal components analysis ofbreeding traits in various ecological regions in Heilongjiang province［J］． Soybean Science，2013，32( 6) : 731-734． )

［2］洪庆红，李丹婷，郝朝运．应用FTIＲ直接测定法鉴定大豆的品种［J］．光谱学与光谱分析，2005，25( 8) : 1246-1249． ( Hong QH，Li D T，Hao C Y． Identification of soybean varieties by directdetermination of FTIＲ spectrum［J］． Spectroscopy and Spectral Analysis，2005，25( 8) : 1246-1249． )

［3］朱大洲，王坤，周光华，等．单粒大豆的近红外光谱特征及品种鉴别研究［J］．光谱学与光谱分析，2010，30 ( 12) : 3217-3221．( Zhu D Z，Wang K，Zhou G H， et al． The NIＲ spectra based varietydiscrimination for single soybean seed［J］． Spectroscopy andSpectral Analysis，2010，30( 12) : 3217-3221． )

［4］杨冬风，朱洪德．基于近红外透射光谱分析和BP 神经网络的大豆品种识别［J］．大豆科学，2013，32( 2) : 249-253． ( Yang DF，Zhu H D．Ｒecognition of soybean varieties based on near infraredtransmittance spectroscopy and BP neural network［J］． SoybeanScience，2013，32( 2) : 249-253． )

［5］柴玉华，侯升飞，彭长禄．基于高光谱图像技术的大豆分级识别方法研究［J］．东北农业大学学报，2014，45 ( 4) : 107-112．( Chai Y H，Hou S F，Peng C L． Identification of different soybeangrades based on hyperspectral imagery［J］． Journal of NortheastAgricultural University，2014，45( 4) : 107-112． )

［6］ Zhang X L，Liu F，He Y，et al． Application of hyperspectral imagingand chemometric calibrations for variety discrimination of678 大豆科学4 期maize seeds［J］． Sensors，2012，12( 12) : 17234-17246．

［7］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 ) : 349-355．( Bao Y D，Chen N，He Y，et al．Ｒapid identification of coffeebean variety by near infrared hyperspectral imaging technology［J］． Optical Precision Engineering，2015，23( 2) : 349-355． )

［8］杨仁杰，刘蓉，杨延荣，等．用二维相关近红外谱和多维主成分分析判别掺杂牛奶［J］．光学精密工程，2014，22 ( 9 ) :2352-2358． ( Yang Ｒ J，Liu Ｒ，Yang Y Ｒ，et al． Classification ofadulterated milk by two-dimensional correlation near-infrared spectroscopyand multi-way principal component analysis［J］． OpticalPrecision Engineering，2014，22( 9) : 2352-2358． )

［9］王斌，薛建新，张淑娟．基于高光谱成像技术的腐烂、病害梨枣检测［J］．农业机械学报，2013，44( S1) : 205-209． ( Wang B，XueJ X，Zhang S J． Detection of decay and disease pear jujube basedon hyperspectral imaging technology［J］． Transactions of the ChineseSociety for Agricultural Machinery，2013，44( S1) : 205-209． )

［10］ Siedliska A，Baranowski P，Mazurek W． Classification models ofbruise and cultivar detection on the basis of hyperspectral imagingdata［J］． Computers and Electronics in Agriculture，2014，106:66-74．

［11］ Nakariyakul S，Casasent D P． Classification of internally damagedalmond nuts using hyperspectral imagery［J］． Journal of Food Engineering，2011，103( 1) : 62-67．

［12］ Sakla W，Chan A， Ji J， et al． An SVDD-based algorithm for targetdetection in hyperspectral imagery［J］． Geoscience and ＲemoteSensing Letters， IEEE，2011，8( 2) : 384-388．

［13］严衍禄，陈斌，朱大洲．近红外光谱分析的原理、技术与应用［M］．北京: 中国轻工业出版社，2013． ( Yan Y L，Chen B ，ZhuD Z． Near infrared spectroscopy-principles， technologies and applications［M］． Beijing: China Light Industry Press，2013． )

［14］ Shahin M A，Symons S J． Detection of Fusarium damaged kernelsin Canada Western Ｒed Spring wheat using visible /near-infraredhyperspectral imaging and principal component analysis［J］． Computersand Electronics in Agriculture，2011，75( 1) : 107-112．

［15］ Jia X，Ｒichards J A． Segmented principal components transformationfor efficient hyperspectral remote-sensing image display andclassification［J］． Geoscience and Ｒemote Sensing，1999，37( 1) :538-542．

［16］ Du Q，Zhu W，Yang H，et al． Segmented principal component analysisfor parallel compression of hyperspectral imagery［J］． Geoscienceand Ｒemote Sensing Letters， IEEE，2009， 6 ( 4 ) :713-717．

［17］ Borges J S，Marcal A Ｒ S，Dias J M B． Evaluation of feature extractionand reduction methods for hyperspectral images［C］/ /Proceedingsof the 26th EAＲSeL Symposium-New Developments andChallenges in Ｒemote Sensing，2007: 255-264．

［18］ Huang G B，Zhu Q Y，Siew C K． Extreme learning machine:Theory and applications［J］． Neurocomputing，2006，70 ( 1 ) :489-501．

［19］ Breiman L．Ｒandom forests［J］． Machine Learning，2001，45( 1) : 5-32．

［20］ Vidal M，Amigo J M． Pre-processing of hyperspectral images． Essentialsteps before image analysis［J］． Chemometrics and IntelligentLaboratory Systems，2012，117: 138-148．

［21］朱启兵，冯朝丽，黄敏，等．基于高光谱图像技术和SVDD 的玉米种子识别［J］．光谱学与光谱分析，2013 ( 2 ) : 517-521．( Zhu Q B，Feng Z L，Huang M， et al． Maize seed identification usinghyperspectral imaging and SVDD algorithm［J］． Spectroscopyand Spectral Analysis，2013 ( 2) : 517-521． )

［22］郭文川，王铭海，谷静思，等．近红外光谱结合极限学习机识别贮藏期的损伤猕猴桃［J］．光学精密工程，2013，21 ( 10) :2720-2727． ( Guo W C，Wang M H，Gu J S， et al． Identification ofbruised kiwifruits during storage by near infrared spectroscopy andextreme learning machine［J］． Optical Precision Engineering，2013，21( 10) : 2720-2727． )

相似文献/References:

[1]黄春燕,刘胜利,王登伟,等.大豆叶面积指数的高光谱估算模型研究[J].大豆科学,2008,27(02):228.[doi:10.11861/j.issn.1000-9841.2008.02.0228]
　HUANG Chun-yan,LIU Sheng-li,WANG Deng-wei,et al.Models for Estimating Soybean Leaf Area Index Using Hyperspectral Data[J].Soybean Science,2008,27(04):228.[doi:10.11861/j.issn.1000-9841.2008.02.0228]
[2]齐波,张宁,赵团结,等.长江中下游生态区大豆生长性状及产量的冠层高光谱偏最小二乘回归预测[J].大豆科学,2015,34(03):414.[doi:10.11861/j.issn.1000-9841.2015.03.0414]
　QI Bo,ZHANG Ning,ZHAO Tuan-jie,et al.Using Canopy Hyperspectral Reflectance to Predict Growth Traits and Seed Yield of Soybeans from Middle and Lower Yangtze Valleys through Partial Least Squares Regression[J].Soybean Science,2015,34(04):414.[doi:10.11861/j.issn.1000-9841.2015.03.0414]
[3]龚荣新,鲁向晖,张海娜,等.基于高光谱植被指数的大豆地上部生物量估算模型研究[J].大豆科学,2023,42(03):352.[doi:10.11861/j.issn.1000-9841.2023.03.0352]

备注/Memo

基金项目: 国家自然科学基金资助项目( 60802059) 黑龙江省自然科学基金重点项目( ZD201303) 。第一作者简介: 刘瑶( 1982-) ，女，博士，讲师，主要从事高光谱图像处理技术研究。E-mail: liuyao0904@163． com。通讯作者: 谢红( 1962-) ，女，教授，博导，主要从事信号与信息处理技术研究。E-mail: xiehong@ hrbeu． edu． cn。

更新日期/Last Update: 2016-08-23