基于主成分分析和主分量分類(lèi)的交通事件自動(dòng)檢測(cè)算法.pdf

1、南京師范大學(xué)碩士學(xué)位論文基于主成分分析和主分量分類(lèi)的交通事件自動(dòng)檢測(cè)算法姓名：武林芝申請(qǐng)學(xué)位級(jí)別：碩士專(zhuān)業(yè)：電子科學(xué)與技術(shù)；電路與系統(tǒng)指導(dǎo)教師：陳淑燕20090517AbstractIIAbstractWiththemodernizationprocessoftransptationtheAID(AtuomaticIncidentDetection)ismemeimptant.Inrecentdecadestransptationres

2、earchershavebeenlookingfimprovethepracticalityofautomaticincidentdetectionalgithm.Thisarticletriestointroduceprincipalcomponentanalysisprincipalcomponentclassificationintoautomaticincidentdetectionarea.Thefourindexstoeva

3、luatethePerfmanceofAIDare:1detectedrate(DR)2flasealarmrate(FAR)3meantimetodetect(MTTD).AsusualwetryourbesttoimprovetheperfermaceofDRMTTDatthepremiseoftheFARisunderthan10%.Principalcomponentanalysistechniquehasbeensuccess

4、fullyusedinimagecompressionintrusiondetectionarea.Assumingthatincidentcasecanbetreatedasoutliersanintrusionpredictivemodelisconstructedbycalculatingthedistancefromthemajminprincipalcomponentsofnmalinstances.Themainpurpos

5、eofusingPCAistoreducethedimentionaryofdatatoextractfeatures.BasedoncaculationingthedistanceinFeatureSpacecomparingthemwecanseparatetheincidentfromthenmal.ThePCCwhichproposedbyW.J.Huasstthetwotypesofsamplesaccdingbymaximi

6、zingthealgebrasumoftwotypessamplesProjectedinthehyperplane.Asusedinincidentdetectionistodivideincidentincidentfree.Inthetrafficdatasetthenumberofincidentcasesismuchsmallwithcomparedtoincidentfreecases.Afewoutliershavelit

7、tleinfluenceontheCategydirectionbecauseitisbasedontheaveragepropertiesofstatisticalsamples.SoPCChasgoodrobustness.ThepaperanalysistheprincipleofPCAOUTLIERPCCthenconstructthemodelofPCAPCC.thesetwomodelsareusedinincidentde