资源描述:
《LandsatTM遥感影像中厚云和阴影去除方法研究》由会员上传分享,免费在线阅读,更多相关内容在工程资料-天天文库。
1、CloudandShadowRemovalfromLandsatTMDataRiPyongSop,MaZhangBao,QiQingWen,LiuGaoHuan(InstituteofGeographicSciencesandNaturalResourcesResearch,CAS,Beijing100101,China)Abstract:Cloudremovalisanimportantstepinremotesensingimageprocess.Inthispaper,theauthorprop
2、osedanewalgorithmforcloudremovalusingmulti-temporalLandsatTMimagedatabasedonspectralcharacteristicsanalysis•Throughthespectralcharacteristicsanalysisofthethickcloudregionanditsshadowregion,thethickcloudanditsshadowidentificationmodelsweredesigned.Usingi
3、mageregression,unsupervisedclassificationandpixelreplacingtechniquesaswellasthesemodels,theinflueneeofthickcloudsanditsshadowscanbeeliminatedorreducedintheLandsatTMimages.Theresultshowsthatthealgorithmcaneliminateorsignificantlyreducethecloudinfluencefi
4、-omLandsatTMimagedata.Keywords:LandsatTMImageData;CloudandShadow;SpectralAnalysis;CloudRemoval1INTRODUCTIONTheearthobservingsatelliteLandsatTM/ETM+rcmotcsensingimagedata,duetoitsenhancedspectralcharacteristics,shortdataacquisitioncycle,widesurveyfield,d
5、atausabilityandotherproperties,havebeenwidelyusedasthemaindatasourceforthestudyofspatial/temporallandusc/covcrchange(LIXia,ctal,1997)・Ithasbeenusedasanidealremotesensingimagedatasourcefortheresearchofregional-scalenaturalresourceandenvironment.However,d
6、uetoclimatereasons,itisdifficulttoobtaincompletelycloudfreeremotesensingimagedata.Mostoftheremotesensingimagedatainclude,moreorless,cloudsanditsshadowsprojectedontheground.Thesegivesometroubletoanumberofusersofremotesensingimagedata.Itbecomesoftenthemos
7、timportantissuehowtoremovetheinfluenceofcloudsfromtheremotesensingimagedata(SongXiaoYu,ctal,2006).So,cloudremovalisanessentialstepintheimagepre-processingprocess(SongXiaoNing,ctal,2003).Agreatnumberofresearchworkcarriedoutontheclouddetectionandremovalsu
8、chasdynamicfilteringmethod(ZhaoZhongMing,1996;WuLu,2003),multi-spectralsynthesismethod,lighttemperaturevaluedifferencemethod,theindexmcthod(SongXiaoNingetal,2003),cloudprocessingalgorithmsbasedonremotesensingimageclassificationre