遙感圖像混合像元信息提取方法研究.pdf

1、分類號密級V938735單位代碼：10019學號：S030257氣蛾差≮犬普’學位論文遙感圖像混合像元信息提取方法研究Extractiontechniquesofmixedpixels’informationforremotesensingimage研究生：指導教師：申請學位類別：專業(yè)領域名稱：研究方向：所在學院：萱墅堂鍪玨熬握壅堂亟攘堂盛業(yè)暹壁攮丕叢基應用疊遮主巫撞堂瞳零零六年六月AbstractRemotesensingincomb

2、inationwithgeographicinformationsystem，globalpositionsystem，allkindsofgroundsurfaceobservingtechnologyandinformationanalysistechnologyandsoondevelopanewkindofearthinformationscience，whichhaveapositiveimpulsefordecisionma

3、kingandmanagenaentofalldifferentlevelsdepartmentRemoteSensingimagehavingimagememberasunitkeeparecord，whichintegratingallkindsofspectralsignalaccordingwithdifferentthingscoveredonthegroundNowremotesensing’squantitativestu

4、dydemandsminimalprocessingunitofcomputerdigitalimageprocessingtransitionfrompixeltosubimageunit，whichpromotesthetotalprecisionofnormalremotesensingclassificationbasedonimagememberThispaperselectedaresearchareaaroundXiaot

5、angshanPrecisionAgriculturalExperimentStation，anddidsomeunmixedimagememherstudyusingthesurfacereflectanceratioofTMMODISimageatthesametimeandspaceThethesisimprovedthenormalmethodsofend—memberabstractingbysupervisedclassif

6、ication，un—supervisedclassification，throughactualspectrummeasurementandfromspectrallibrariesThismethodgotspectralinformationbyusingMNFandPPIfromTM，verificatedsimilaritybetweentheresultsandactualspectruminformationbasedon

7、shapefactorwhichdeclaredthatthespectrumdatumcouldreplaceactualspectrummeasurementThestudyutilizedthepureend—memberspectrumderivedfromTMtodecomposethemixedpixelsofTMandMODtSandgotthe15％fractionalerrorreferringtothearea’ss

8、tatisticalinformationoflandusedatabasein2004andfindthedecompositionofTMimagewasmoreprecicenSothestudyshowedthatitwasfeasibleandeffectivetousehighresolutionimagetogettheendmemberfrommultispectrumremotesensingimageanddistr