21462.基于面向?qū)ο蟮母叻直媛蔬b感影像道路提取研究

1、分類號：密級：學(xué)號：2009263單位代碼：10407碩士學(xué)位論文論文題目:基于基于面向?qū)ο竺嫦驅(qū)ο蟮母叻直媛蔬b感高分辨率遙感影像道路影像道路提取研究提取研究研究方向遙感遙感技術(shù)與應(yīng)用技術(shù)與應(yīng)用專業(yè)名稱大地測量學(xué)與測量工程大地測量學(xué)與測量工程研究生姓名申真導(dǎo)師姓名、職稱劉小生劉小生教授教授20122012年6月3日江西贛州IIAbstractTheroadistheimptantfeaturesinhighresolutionremot

2、esensingimageconstitutestheskeletonofthecityisoneofthefastestupdatefundamentalgeographicinfmationofthehighspeeddevelopmentofthecity.Therearemanyalgithmsftheroadextractionhoweverfhighresolutionremotesensingimageoftheroade

3、xtractionalthoughscholarshavestudiedalotbutinthepresentresearchthestudiesarestillintheexperimentaltheeticalstagenotfullyappliedtoproductionpractice.Therefethestudyontheroadextractionfromhighresolutionremotesensingimagesh

4、asacademicsignificancepracticalvalue.Thispaperusesobjectientedimageanalysismethodtoresearchthemethodoftheroadextractiononthedifferentscaleofhighresolutionremotesensingimage.Firstlyonthebasisoftheeticalresearchthepaperstu

5、diedtheobjectientedimageanalysisofscaletransfmingscaleeffectsegmentationparameterion.Accdingtotheimageinfmationlossdegreeintheprocessofscaletransfmingofexperimentalevaluationthepapertestverifythescaleeffectwhichisledbyth

6、edifferentsegmentationscaledeterminetheoptimalparametersonmultiscalesegmentationproposedakindofmethodofoptimalsegmentationscalebasedonthetrainingsampleidentifytheoptimalsegmentationscaleondifferentroadtypesSecondlythepap

7、eranalysisthedifferentroadfeaturesoftheresearchareaintroduceakindofvisualizationoffeatureionmethodtoachieveeffectivecombinationfeaturescreatetheimageobjecthierarchythroughtheroadscaleconversionion.Combiningthetwoclassifi

8、cationmethodofnearestneighbfuzzylogicthepaperoptimizethefeaturespaceconstructtheknowledgebaseoftheroadsidentifytheroadwkThirdlyusingthemathematicalmphologythepapergetthewholeroadwkinthepostprocessingFinallythepaperevalua

9、tethemultiscaleroadextractionaccuracybyusingtheclassificationofstabilitytheconfusionmatrixcomparethemethodwiththeotheranalyzethecontrastiveresult.Theexperimentalresultsshowthatthemethodofobjectientedmultiscaleroadextract

10、ioncaneffectivelysolvethedifficultiesinthehighresolutionremotesensingimageroadextractionfullyembodiesthefeasibilityefficiencyofthisroadextractionmethod.Keywds:highresolutionobjectientedroadextractionscaleroadknowledgebas