流行學(xué)習(xí)在交通標(biāo)志識別中的應(yīng)用研究.pdf

1、西安電子科技大學(xué)碩士學(xué)位論文流行學(xué)習(xí)在交通標(biāo)志識別中的應(yīng)用研究姓名：李福才申請學(xué)位級別：碩士專業(yè)：計算機應(yīng)用技術(shù)指導(dǎo)教師：方敏20100101AbstractAbstract?Thetrafficsignsrecognitiontakesanimptantresearchareaoftheintelligencetransptationsystem.Thefunctionsofitintheaspectsofroadsafetyinfm

2、ationdirectionsoonwillbecomememeimptant.Butthecurrenttrafficsignrecognitioninthedataamountrecognitionrateotheraspectsisnotsowell.Someeffectivefeatureextractionmethodsmayleadagoodideatosolvethisproblem.Manifoldlearningmet

3、hodscanexpletheintrinsicrelationshipbetweenthedataeffectivelyreducethedimensionsofthedata.BygivingaresearchofthetypicalalgithmofmanifoldlearninginthispaperweexperimentontheSwissRolldatamodeluseittocamparethelineardimensi

4、onreductionmethodswiththenonlinearonespointouttheadvantagesoflocallylinearembedding(LLE)methodsinthedatadimensionalityreduction.IndertotestusetheLLEdimensionalityreductionmethodtothetrafficsignrecognitionlaterfirstthepub

5、licfacedatabaseLisusedonthefacerecognitionexperiment.BycomparingtheLLEwithPCALDAmethodsLLEmethodhasagoodadvantageofdiscoveringthelowdimensionalfeaturesintheimage.Theadvantagesthechoiceofitsparametersthroughexperimentsare

6、discussed.ThenthedirectionalsignsaredisposedtoverifythatLLEmethodisfeasibleinthetrafficsignrecognitionapplications.Intheapplicationresearchoftrafficsignrecognitionfirstlysoftwareframewkofthetrafficsignrecognitionisdesign

7、ned.Withthecolshapefeaturesofroadiconswemakeuseofthemomentinvariantstodetectthetrafficsigns.ExtractingtheeffectivelowdimensionalfeatureoftrafficsignimageswithLLEmethodusingAdaBoostensemblelearningalgithmwithBPneuralwkcla