基于高光谱图像处理的大豆品种识别-论文.pdf

《基于高光谱图像处理的大豆品种识别-论文.pdf》由会员上传分享，免费在线阅读，更多相关内容在行业资料-天天文库。

1、第30卷第9期农业工程学报V_0l_30No．920145月TransactionsoftheChineseSocietyofAgriculturalEngineeringMay2012235IdentificationofsoybeanseedvarietiesbasedonhyperspectralimageTanKezhu

2、ChaiYuhualSongWeixianICaoXiaoda(CollegeofElectricalandInformation，NortheastAgriculturalUniversity,Harbin150030，C

3、hina)Abstract：Diferentsoybeanseedvarietieshavediferentcomponents(oil，protein，fatetc．)content．Identificationofsoybeanseedvarietiesisacriticalfactorthatimprovesthequalityofproducedsoybear1．Inthisstudy,hyperspectralimagetechniquewasappliedinordertoclassifysoybeanseedsbasedontheirv

4、arieties．Thespectralreflectancedatawascollectedusingtheopticalsensorsystemwithspectralregionof1000—2500nm．Principalcomponentanalysis(PCA)wasperformedtoreducethedimensionalityofthedataandremovetheredundancy．ScoresoffourPCswereusedasinputfeaturesintheclassificationalgorithm．Fourt

5、exturefeatureparameters(angularsecondmoment，energy,entropyandcorrelation)wereextractedfromeachfeatureimageselectedbyPCA．Fortheextractionofspecificfeatures，foursignificantfeatureparameterswerecomputedfromthe16characteristicvariables．Artificia1neuralnetworkfANN)classifierwasemplo

6、yedforclassificationusingtopselectedfeatures．Theobtainedaveragetrainingaccuracyratewas97．50％andtheaveragetestingaccuracyratewas93．88％．Thus，theresultsconfirmedthathyperspectralimagetechniquein．conjunctionwithBPneuralnetworkcouldbeusefulforsoybeanseedvarietiesclassification．Keywo

7、rds：hyperspectralimage；imagingprocessing；principalcomponentanalysis；soybeanseedclassification；texturalfeature；BPneuralnetworkdoi：10．3969／i．issn．1002—6819．2014．09．029CLCnumber：0657．33；$651Documentcode：AArticleID：1002—6819(2014)～09—0235—08TanKezhu，ChaiYuhua，SongWeixian，eta1．Ident

8、ificationofsoybeanseedvarietiesbasedonhyperspectralimage[J]．TransactionsoftheChineseSocietyofAgriculturalEngineering(TransactionsoftheCSAE)，2014，3O(9)：235—242．(inEnglishwithChineseabstract)谭克竹，柴玉华，宋伟先，等．基于高光谱图像处理的大豆品种识别[J]．农业工程学报，2014，30(9)：235—242．soybeans，advancingmarketcompe

9、titivenessby0Intr0ductiOnclassifyingsoybeanseedsvariet