一种基于椭圆肤色模型的人脸检测(参考价值大)方法

170.2006.14(2)ComputerMeasurement&Control:2005-07-08;:2005-08-19:(2004SM006):(1979-),,,,DSP,:1671-4598(2006)02-0170-02:TP274:B,(,100044):,YCbCr,BP;,:robust;:;;FaceDetectionUsingEllipseSkinModelLiJie,HaoXiaoli(ElectronicandInformationSchool,BeijingJiaotongUniversity,Beijing100044,China)Abstract:AnewfacedetectionmethodispresentedbybuildingupaskinmodelspaceofYcbCrasapreprocessingway.Thealgorithmisrobusttodifferentlights,sheltersandorientation.Experimentalresultsdemonstratetheefficiencyofthisalgorithm.Keywords:wavelettransform;BPNetwork;Lightingcompensation20(Facedetection),,:,,:[1][2][3][4](SVM)[5](WT)[6](HMM)2001Sheng-funLin[7],,HaarBP111,,,,,,,,,:RGBRGBHSV(HIS)YCrCbYIQYESCIEXYZLUVY,,5%(),,RGB255,:(a);(b),,Rein-LienHsu[8]YCbCrRein-LienHsuCbCr,Y2(1)Cb-Cr,2(2)YCbCr,3(Y),(Cb/Y)-(Cr/Y)2,:1713(Cb/Y)-(Cr/Y):Ci(Y)=(Ci(Y)-Ci(Y))WCiWCi(Y)+Ci(Kh)ifYKlorKhYCi(Y)ifY[Kl,Kh](1)WCi(Y)=WICi+(Y-Ymin)(WCi-WLCi)Kl-YminifYKlWHCi+(Ymax-Y)(WCi-WHCi)Ymax-KhifKhY(2)Cb(Y)=108+(Kl-Y)(118-108)Kl-YminifYKl108+(Y-Kh)(118-108)Ymax-KhifKhY(3)Cr(Y)=154-(Kl-Y)(154-144)Kl-YminifYKl154+(Y-Kh)(154-132)Ymax-KhifKhY(4),(1)CiCbCrMIT,:Wcb=46.97,WLcb=23,WHcb=14,Wcr=38176,WLcr=20,WHcr=10,Kl=125,Kh=188YCbCr,Ymin=16,Ymax=2354Y-CbY-Cr,Ci(Y)WCi(Y)/245Cb-Cr(5)(6):(x-ecx)2a2+(y-ecy)2b2=1(5)xy=cos,sin-sin,cosCb-cxCr-cy(6),cx=109138,cy=152102,=2153(rad),ecx=1160,ecy=2141,a=25139,b=14103,Cb-Cr5YCbCr,1400341613462,Haar,BP(182)1822,DataSocketdstp,URL:dstp://ip/dataAPI,,412DataSocketAPI41211VILabVIEW711ActiveXEventCallbackforIE1viVIHtmlVI,ActiveXEventCallbackforIE.viIELabVIEW,LabVIEW741212DataSocketAPIDataSocketAPIDataSocketServerVIVIDataSocketVI85,,Lab2VIEWRunTimeEngine,,IEIP,,HtmlVIIP,WebServer,DataScket,Internet,,,LabVIEW:[1].LabVIEW7Express[M].:,2004.[2],,.[M].:,2004.[3],.LabVIEW[M].:,2001.[4].[M].:,2003.[5],.[M].:,2001.[6].[M].:,2001.[7],,.CADOrCAD912[M].:,2003.(171)7:3/3;5/5;12/103MIT,20320,Haar,64BP1.2,,robust,222494249299.95%5779577799197%74,HaarBP,,,,,Integralimage[9]:[1]CaiJ,GoshtasbyA.Detectinghumanfacesincolorimages[J].ImageandVisionComputing,1999,18(1):63-75.[2]YangGZ,HuangTS.Humanfacedetectioninacomplexback2ground[J].PatternRecognition,1994,27(1):53-63.[3]BrunelliR,PoggioT.Facerecognition:featuresversustemplates[J].IEEE:TransPatternAnalysisandMachineIntelligence,1993,15(10):1042-1052.[4]RowleyHA,BalujaS,KanadeT.Neutralnetwork-basedfacedetection[J].IEEETransPatternAnalysisandMachineIntelli2gence,1998,20(1):23-38.[5],.[J].,2003,11(12):914-917.[6],,,.[J]2003,1.[7]LinSF,ChenJY,ChaoHX.Estimationofnumberofpeopleincrowdedscenesusingperspectivetransformation[J].IEEETrans2actionsonSystems,ManandCybernetics-PartA:SystemsandHumans,2001,31(6):645-653.[8]HsuRL,etal.Facedetectionincolorimages[J].IEEETrans.onPatternAnalysisandMachineIntelligence,2002,24(5):696-706.[9]ViolaP,JonesM.Rapidobjectdetectionusingaboostedcascadeofsimplefeatures[A].Proceedingsofthe2001IEEEComputerSoci2etyConferenceonComputerVisionandPatternRecognition,2001[C].2001,2001:511-518.