用于高光谱图像分类的归一化光谱指数的构建与应用-论文.pdf

《用于高光谱图像分类的归一化光谱指数的构建与应用-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43卷第2期红外与激光工程2014年2月Vo1．43No．2InfraredandLaserEngineeringFeb．2014DevelopmentandapplicationofnormalizedspectralindexbasedonhyperspectralimageryclassificationZhangDongyan，ZhaoJinling一，HuangLinsheng一，MaWenqiu。(1．KeyLaboratoryofIntelligentComputing＆SignalProcessin

2、g，MinistryofEducation，Hefei230039，China2．BeijingResearchCenterforInformationTechnologyinAgriculture，Beijing100097，China)Abstract：Near——groundimagingspectroscopyappliedinfieldprovidesnewopportunityfordevelopmentofquantitativeremotesensinginagriculture．Itdeserves

3、concemabouthowtoexertitsdataadvantageofintegratingimageandspectraintoone，particularlyinanalyzingtheinfluenceofbackgroundtargets，suchassoil，shadowoncropnutrientinversionmode1．Inthisresearch，imagingcubesofwheatgroupinthefieldwerecollectedbyvisible／near-infraredim

4、agingspectrometer(VNIS)．Anormalizedspectralindexwassetupaccordingtoreflectancecharacteristicsofilluminatedsoil，shadowsoil，illuminatedleafandshadowleafintheimage．Furthermore，theindexwasusedtoextractspectraofdifferenttargetsinsoybeanimagesandanalyzethevariationof

5、determinationcoeficientRbetweennormalizedspectraofsoybeangroupandchlorophylldensitybeforeandafterremovingbackgroundsoil．Theresultsshowedthatwhenspectraofsoilandshadowleafwereremoved，thesensitivebandsofchlorophylldensityshiftedfromredandnear—infraredranges(727nm

6、，922nm)toredranges(710nm，711nm)，meanwhile，theoveralltrendwasthatvisiblerangesincreased，near-infraredregionsdecreasedandredbandshadthehighestdeterminationcoeficient．Therefore，itcanbeconcludedthatspectralpurificationbasedonnormalizedspectralindexhasimportantsigni

7、ficanceforquantitativeresearchinagriculturalremotesensing．Keywords：hyperspectralimaging；normalizedspectralindex；imageryclassification；spectrumpurification；wheat；soybeanCLCnumber：S127Documentcode：AArticleID：1007—2276(2014)02—0586—09用于高光谱图像分类的归一化光谱指数的构建与应用张东彦1,2，

8、赵晋陵1,2，黄林生1,2，马雯蔌(1．安徽大学智能计算与信号处理教育部重点实验室，安徽合肥230039；2．北京农业信息技术研究中心，北京100097)摘要：成像高光谱的近地田间应用为农业定量遥感的发展提供了新的契机。如何发挥其图谱合一的数据优势，尤其在解析土壤、阴影等背景地物对作物养分反演模型的影响需要关注。该研究借助可见／近红外收稿日期：