基于LabVIEW和机器视觉铆钉表面缺陷在线检测系统研究

iiiLabVIEWcostarSI-M350NIPCI-1410LEDLabVIEWLabVIEWNIIMAQNIVisionivvRESEARCHOFONLINERIVETSSURFACEDEFECTDETECTIONSYSTEMBASEDONLABVIEWANDMACHINEVISIONModernindustriesaskforcontinuallyincreasingdemandsofthequalityoftheworkpiece.BlindrivetswithbreakpullmandrelusedtooccasionswhichisinconveniencewithordinaryRivetriveting(rivetingfromtwosides),widelyusedinconstruction,automotive,shipbuilding,aircraft,machinery,electricalequipment,furnitureandotherproducts.Thequalityhascrucialinfluencetothestrengthoftheproduct,sportsperformance,servicelife,andsoon.However,assomereasonsinproductionprocess,therewillbesomerivetssurfacedefects,suchascracksandfissures,whichhaveseriouslyaffectedthequalityofproducts.Sowehavetomakeastringentinspectionofthequalityofrivets.ThemaintopicofthisresearchistodevelopaRivetsurfacedefectdetectionsystembasedonmachinevisionwhichisforrivet-orientedenterpriseapplications.Tosetupthehardwareplatform,thesystememployscostarSI-M350industrialautomationcamera,NIPCI-1410Imageacquisitioncard,andself-madeLEDlightsource,industrialcomputerandsoon.Aboutsoftware,mainlyonLabVIEWvirtualinstrumentsandrelatedknowledge,aswellastherealizationofrivetssurfacedefectdetectionsystemonit.Achieveaseamlessconnectivitywithhardware.viAfterpractice,thesystemisstable,rapidandaccurate,satisfyingtheindustrialscenerivetssurfacedefectdetectionrequirements.Throughimageprocessingforindustrialinspection,willimprovetheefficiencyandthequalityofdetection.Ultimatelyimprovebusinessefficiencyandcompetitivenessofthecompany.Therefore,imageprocessingindustrialapplications,hasbroadmarketprospectsandresearchvalue.KEYWORDS:machinevision,virtualmachines,onlinetesting,surfacedefects652008-1-25662008-1-252008-1-2511.11-1Fig.1-1Rivetanditssurfacedefects2,1.280%[1,2][3]31.2.1205060Roberts205060Roberts70MIT(MITAI)DavidMarr802070CCDCCD2080CPUDSP20802090[9]1.2.21-2[3]41-2Fig.1-2Structureofmachinevisionsystemmodules1.2.31-3[4]/5/1-3Fig.1-3Typicalmachinevisionsystemapplicationsdiagram1.361PLC2USB341.41.4.1123000345%71.4.211203000234LabVIEWLabVIEW82NILabVIEWNIvisioncostarSI-M350NIPCI-1410LED2.12.1.12-1[5]2-1Fig.2-2Rivetmechanicalperformancelevels92.1.2[6]1()2342-2102-2Fig.2-3RivetSurfaceDefects2.1.310000GB/T3098.19-2004GB/T3098.18-2004/ISO14589:2000[7]ISO6157-1;1988124mm1mm4mm230.05mm40.05mm5120/3000/697%95%2.1.4120112-32-3Fig.2-3SystemflowchartPCI-1410PCILabVIEWcontrolPLC2.2costarSI-M350NIPCI-1410LED2-4122-4Fig.2-4Hardwaresystemcomponents2.2.11CCD/CMOS[8]CCD(ChargeCoupledDevice)CCDCCDCCDCCDCCDMOSMOSCCDPCCDCCD13CMOSComplementaryMetal-OxideSemiconductorCCDCMOSCMOSN–P+CMOS,CMOSCCDCCDCCDCMOS2-5(a)CCD(b)CMOS2-5CCDCMOSFig.2-5CCDandCMOS2CMOSCCDCMOSCCDCMOSCCDCCDCMOSCMOSCCD1/3CMOS[9]CMOSCCD300CCD300CMOSCMOS14CCDCostarSI-M3502-62-6CostarSI-M350Fig.2-6CostarcameraSI-M35011/2HyperHADITCCD2EIA:768()4943CCIR:7525824556dB67(Field)(Frame)89VD10OEMFlextrig11TTLHD/VD12EEN(WEN)131415C-mount15SI-M3501/1000s//[10]3Computar11/2C12-120mmF1.8-560C212-120mm3F1.8-560C47079.5123.5mmComputarCostar12cm42-72-7Fig.2-7Camerasystem2.2.2ComputarCostar161234:[11]LED130KHz2()3174LEDLEDLighyEmittingDiode[12]LED1LEDLED2LED30000103LED45LED2-82-8Fig.2-8Lamp-housedesign18[13]LED2-92-9Fig.2-9Lamp-houseirradiation2.2.3NIPCI-1410NI2-10PCI-1410192-10PCI-1410Fig.2-10PCI-1410imageacquisitioncardNIPCI-1410[14]PCI-1410IMAQPCI-1410UserManual2.2.495Keyence-,20dark2-112-11Fig.2-11Sensoranditsinstallation2.2.512TTLTTL2-122-132-122-13Fig.2-12JunctionboxoutsideFig.2-13Junctionboxinside212.2.61CPUP43.0GHZ512M128M80G2123IMAQ-68222-142-14IMAQ-6822Fig.2-14IMAQ-68222.32.3.122NINIVisionBuilderAIVisionAssistantVisionBuilderAssistantLabVIEWC[15]NINIVisionAssistantLabVIEWIMAQ--NIVisionAssistantIMAQNIVisionAssistantLabVIEWNIVisionAssistant[16]12382(Blobanalysis)45NIVision8.010[17]NIIMAQVisionBuilderIMAQVisionIMAQVisionBuilderVI[18-22]IMAQVision8.0400NIIMAQ23IMAQVision8.02.3.2NIVisionassistant8.0NIVisionassistant8.0LabVIEWVIC[23]NIVisionassistant8.02-152-15NIVisionassistant8.0Fig.2-15NIVisionassistant8.0userinterface2.3.3LabVIEW8.0Visionassistant8.02-16242-16Fig.2-16Softwarestructure2.4LabVIEW8.0NIvision25383.13-13-1Fig.3-1Typicalrivetsurfacedefectsimage3-13-2351100152615353-2Fig.3-2Typicalrivetsurfacedefectsimagegraydistribution3.23.2.1CCD[23]12100CCD10010[24-28]3.2.227FFT1[NN]G(i,j)CMN28∑∈=CjijiGNNjiG),(),(*1),('(3-1)G’(i,j)G(i,j)N333-3(a)3-3(b)11111111191131353131211ab3-3Fig.3-3Selectionofmeanfiltertemplate2G(i,j)G’(i,j)333-43-53-6293-43-53-6Fig.3-4OriginalimageFig.3-5AftermeanfilterFig.3-6Aftermiddlefilter3.2.3GminGmax[Lmin,Lmax]LGiG’i:maxminmaxmin'iiLLGGGG−=−3-23-73-8303-73-8Fig.3-7OriginalFig.3-8Afterimageenhancement3.31SobelRoberts23313.3.1[I(w,h)]WH,WH80255kAB0knAk255nBnA+nB=M×NµµAµBωAωB:==∑∑∈∈BBnmIBAAnmIAnhmInhmI),(),(),(),(µµ3-3×=×=NMnNMnBBAAωω(3-4):BBAAµωµωµ+=3-5ωA+ωB=1()()()()2222ABBABBAAkµµωωµµωµµωσ−=−+−=3-6nAnB()()22ABBAnnkµµσ−==3-72(k)kTh()(){}kThk225502maxσσ≤≤=3-83-9323-9Fig.3-9Afterthreshold3.3.23-94ThW1flag[WH]WH0102331S1441S15536SSThW3SThW33ThW3-103-10Fig.3-11Eliminatethein