公交车辆行驶安全智能监控系统开发与设计

上海交通大学硕士学位论文公交车辆行驶安全智能监控系统开发与设计姓名：阳衡申请学位级别：硕士专业：机械电子工程指导教师：刘成良20080201I863ZemikeHoughIIABSTRACTIIIDevelopmentofIntelligentPublicTransportSafetyDrivingMonitorSystemABSTRACTTrafficsafetyisaperpetualthemeinthedevelopmentofautomobiletraffic,withtherapidincreaseofvehiclepopulation,roadtrafficaccidents,particularlythehighincidenceofseriousaccidents;safetyproblemshavebecomeincreasinglyprominent.Thepassengersofbusorcoachwillbeinjuredbadlyinthevehicleaccidentbecausetherearealwaystoomanypassengersonthevehicle.Soitisveryimportantforustoimproveanddevelopvehicleespeciallythepublictrafficvehiclessecurityguardtechnology.Thisarticleembarksfromthevehicleactivesafety,hascarriedonsomepositivebeneficialexplorationsinthevehiclessafetyensureareaoftechnology.BasedonthekeytechnologyoftheNationalProgram863”Researchontheintelligentnavigationsystemoffarmmachinebasedoninformationfusionofmutlisensors”,somespecialimprovementhavebeenmadeaccordingtothedifferentresearchobjectofthisarticle.Itisnotdifficulttoextractsimpleinformationsuchasroadlanemarkandtherelativepositionofvehicleontheroadfromthemachinevisionsensor,sowehavemadesomeimprovementonthelanemarkidentificationandtrackingtechnologyincludingusingamethodbasedonZernikeMomentlinedetectionalgorithm,andthenuselanemarkfeaturestoextractthelanemarkingfeaturepoint,anduselinearregressionmethodtoextractlanemarkparameters.Houghtransformwasusedtoextractthelanemarkparametersthroughmappingthelineofimageinterspacetonavigationparametersinterspace.AccordingtotheimagelanemarkingparameterinformationnotthepositionanddirectionoftheparameterestimationmodelinthecurrenttrafficABSTRACTIVlanewereestablished,butalsotheestablishmentofalanewidthofthemodel.Thesemodelscanbeusedtoestimatethecurrentparametersoftheposition,directionandlanewidthincurrentvehiclelanes,andmaketheLaneDepartureWarningpossible.Thearrivaltimepredictionmethodofintelligentvehicleadvancedinthisarticleisbasedonunarylinearregression.Thoughthismethod,theeffectfactoroftrafficconditionwasdrewvieunarylinearregressionanalysis,andthedownstreamvelocitycanbeforecastedaccordingtothecurrentvelocityofvehicle.Theforecastingaccuracyofvehicle’sarrivingtimeimprovedbecausemanyfactorssuchasarrivaltimehadbeenconsidered.In-vehiclesecuritymonitoringandcontroldevicewasdeveloped,andthisdeviceisprovidedwiththefunctionofLaneDepartureWarningandarrivaltimeprediction.Meanwhile,Localinformationdisplayequipmentwiththefunctionofdisplayingarrivaltimeandthevehicle’sreal-timepositionwasdeveloped.TheresearchgoalistodeveloponekindofLaneDepartureWarningSystembasicallydrawingclosetothepracticalnessandprovidethetheoreticalandtechnicalsupportforourcountryAutomotiveSafetyDrivingAssistSystem.Keywords:IntelligentTransportSystem,machinevision,LaneDepartureWarning,arrivaltimeprediction20083520083520083511.12003116220038.618ESPElectronicStabiltyProgram[1]2ESP1.2,,,,:1,,,,,,PAPRICA,,,LED,,LED[2]IterisAutoVueTM[3],,ValeoValeos,Iteris,MobileyeN.V.EyeQEyeQ22,,3,,,MOBOLAB18Maxi,1CCD,4LED:;;;,[4]2CCD,,,,,,,DEMO[5],,,3,,,[6]4,[7],,,,,,YasuhiroSuzuki[8],,Chan,[9],,,4,YitingLiu3[10]20011.3IVITS2004567753992174518101942926228005601901800040LDWSLDWS[11-17]1.4GPS/GPRS51-11-1Fig.1-1PrincipleofIntelligentPublicTransportSafetyDrivingMonitorSystem6ARM7C/OS-GPRS23272.1[18-19]:;[20]2-12-1Fig.2-1StructureofLaneDepartureWarningsystemARMUSBCOM82-22-2Fig.2-2FunctionofLaneDepartureWarningsystem2.22-32-3Fig.2-3PositionbetweencameraandroadL=5mH=2.2m9=51°222tan5.22WHLmα=×+×≈0arctan()11.3HLθ=÷≈L2.3[21-23]2-4Fig.2-4PositionofBusonthecurrentLane2-4yPyxxyO100xPxPy0xPy0xy0xCCD2-10010cos2bxxθθθθ+×≥ff2-12-20010cos2bxxθθθθ−×≤−pf2-2b0θ11xy1x2-312002wx=−2-31x2-411002wx=+2-4w2-5112-5;2-5Fig.2-5PrincipleofLaneDepartureWarning2.3CCDCCD2-6122-6Fig.2-6DynamicFrameofBusyyxxyOEFd;;MNCCDhMNyM00(,)llkb00(,)rrkb(_,_)kleftbleft(_,_)krightbrightCCDCCD0xθw2.3.1(2-5)__180ar2nkleftkrightctgkθπ+=−××2-513knCCDkn(2-6)()()20__2xnHimagekfWimageyµ×=×+×2-6_Himage_Wimage;fCCD;xµCCD;0y2-7xoyy(-15)2-7Fig.2-7FrameofCamera2.3.2CCD()()λ(2-7)λ/140xyhyµλµ×=×2-7(2-7)xµyµCCD;hCCD;0yCCDxµ=yµ2-80hyλ=2-8λθ(2-9)w__cos__bleftbrightwkleftkrightλθ=×−×2-92.3.30xQMQMQM(2-10)__(_)__2brightbleftQMWimagekrightkleftλ=−−×2-10(2-8)_Wimage;λ(2-7)/QMθ(2-11)0x0cossinxQMhθθ=×−×2-112.3.415,,[24]2-82-8Fig.2-8LDWarithmeticchart2.4NoYesCCD16RGB243.1RGB24RGB256(28)16777216256*256*256RGB256[25]RGBYUVYRGBYUVYRGBY=0.3R+0.59G+0.11B3-110x4173-1Fig.3-1Bargraphofthefigure25602552542563-23-2Fig.3-2ExampleofBargraph[26]3-3183-3Fig.3-3Double-peaksfeatureofbargraphY=cr*R+cg*G+cb*Bcr+cg+cb=1cr0cg0cb03-2crcgcbRGB//foreachlinefor(inty=0;yheight;y++){//foreachpixelfor(intx=0;xwidth;x++,src+=3,dst++){*dst=(byte)(cr*src[RGB.R]+cg*src[RGB.G]+cb*src[RGB.B]);}src+=srcOffset;dst+=dstOffset;}3.23.2.101019255Y1255003-4[27]3-4Fig.3-4Imagebinaryzation3.2.220[28](1)p-1962Doylep-(p-tile)PO/PB,Tf(x,y)=Tf(x,y)T(2)T0T0T0TAT0TBTATBT1T1T0TKTK1TK(3)Otsu’smethodOtsu1978(4)ShannonPunKapur21ttttPpPpPp,,,10ΚtlttttPpPpPp−−−−++1,,1,1121Κ3-3(),log02tititiBPpPptH∑=−=()tiltitiWPpPptH−−−=∑−+=1log