a knowledge-based appoach for reducing cloud and shadow

《a knowledge-based appoach for reducing cloud and shadow》由会员上传分享，免费在线阅读，更多相关内容在教育资源-天天文库。

1、AKnowledge-basedApproachforReducingCloudandShadowMingjunSongandDanielL.CivcoLaboratoryforEarthResourcesInformationSystems(LERIS)DepartmentofNaturalResourcesManagement&EngineeringTheUniversityofConnecticutU-4087,1376StorrsRoadStorrs,CT06269-4087ProblemCompletelycloud-freeremotelysensedimages

2、arenotalwaysavailable,especiallyintropical,neo-tropical,orhumidclimates,posingcomplicationsandperhapsseriousconstraintstoimageanalysisObjectiveDevelopaknowledge-basedmethodtoproduceacloudandcloudshadow-freemultitemporalimagecompositeofneo-contemporaryimagesMethodologyTopographicalnormalizat

3、ionMulti-dateBrightnesscorrectionMainimageversussecondaryimageTodetectthoseareaswhicharecoveredwithcloudandshadowinthemainimage,andnothavingcloudorshadowinthesecondaryimageReplaceFirstStudyAreaEasternMadagascarDecember15,2000April22,2001TopographicalNormalizationNormalizationResultMulti-dat

4、eBrightnessCorrectionCriteriaBand1:CloudBand4:ShadowBandDifferenceRationale:Ifthedifferenceislessthanathreshold,itshouldbethesameobjectintwodatesimagesCriteria:ShapeInformationShadowandstream.eCognition,length/widthLength/width:BoundingBox:CovarianceMatrix:KnowledgeBaseParameterCloudShadowO

5、utputValue21Band1(MainImage)>41>=0Band4(MainImage)>=0<35Band1(SecondaryImage)<33<33Band4(SecondaryImage)>=0>27Band1Difference>10>=0Band4Difference>=0>10Length-to-WidthRatio>=0<9ImagineSpatialModelCloudandShadowDetectionExpert01_band100_band101_band400_band4Band1_difBand4_difObj_lwCompositeR

6、eplaceSecondStudyAreaETM23April2001ETM26March2000Central-EasternConnecticutThamesRiverUConnMulti-dateBrightnessCorrectionSample_01Sample_00Sample_00_corr00_corrCriteriaBand1:CloudBand4:ShadowBandDifferenceContextualInformationDifficulty:CloudedgewithurbanareaShadowwithwaterareaRationale:Clo

7、udedgeandshadowareashouldbeaccompaniedbyclearcloudareaBuffertheclearcloudareaKnowledgeBaseParameterCloudShadowOutputValue21Band1(MainImage)>95>=0Band4(MainImage)>=0<46Band1(SecondaryImage)<221<221Band4(SecondaryImage)>=0>=0Band1Difference>10>=0Band4Diffe