机器视觉在RFID标签封装设备中的应用研究

华中科技大学硕士学位论文机器视觉在RFID标签封装设备中的应用研究姓名：刘少勇申请学位级别：硕士专业：机械电子工程指导教师：熊有伦;王瑜辉20070529IRFIDRFIDXY20µm1RFIDIIAbstractAsICscontinuetobecomemoresophisticatedandminimumfeaturesizeskeepshrinking,theneedforhigheraccuracyinfinepitchinterconnectapplicationshasbecomemoreimportantthenever.Machinevisioncanidentifyandlocatethemicrochipsthroughopticalsystemdesignandimageprocessingtechnology,soitiswidelyusedinsemiconductorpackagingequipment.Forsmallersize,higheraccuracyandhigherspeedforRFIDtagpackageequipment,thehardwareimplementandimageprocessingofvisionalignmentsystemaremorecomplexthanever.TheresearchworkofmypaperisbasedonRFIDtagpackageequipmentswhichisdevelopedindependentlybyourgroup.Theimplementofon-the-flyvisionalignmentsystemandcorrespondingimageprocessingtechnologyofequipmentsarestudiedsystematically.Themaincontentsofmypaperareasfollows:Multiimageprocessingmethodsarestudiedtoenhancetheadaptabilityandflexibilityofvisionsystem.Firstly,smoothing-filteringapproachestothefiducialmarkareintroducedinmypapercontainmedianfiltering.Secondly,thresholdsegmentationapproachestotheimagecontainsimpleiteratingmethodofoptimumthresholdsegmentationandmaximum-minimumvariancessegmentationmethods.Anautothresholdalgorithmisdesignedtoseparatethefeaturefromthebackground.Finallytwomethodsofnormalizedgrayscalecorrelationarecompared.Forfailingtolocatetheimageformissingfeaturesofthefiducialmark,anefficientapproachtotheimagelocatingisdesignedforit.Amultiplereflectorvisionsystemschemeisdesignedtomeettherequirementsofhighspeedmountingofthecomponents.Calculationsforbiasandrotorcenterofbonderheaderarediscussedindetails.Thispartalsoanalyzesspeedandaccuracyofthesystem.Experimentalresultsdemonstratethattheycanmeettheneedofhigheraccuracyandreal-timeperformance.Basedontheworkingprincipleoftheequipment,thepaperdiscussestheworkingflowofthreesetsofvisionsystems.Thehardwareconstructionandsoftwaredevelopmentofmultivisionsystemsareintroducedindetail.Consideringallocationofsystemresources,themultithreadtechnologyisproposedinvisiondevelopment.Thecontentofthepaper,canbeastheresearchinformationofthemachinevision’sapplicationinthepackagingofthesemiconductor,andcanbethereferenceasthepartofmachinevisioninotherrobotizedproductions.Keywords:RadioFrequencyIdentification;GrayscaleEnhanced;ThresholdSegmentation;VisionAlignment;MultiThread;MissingFeature______11.1.19732003CB71620750390063863RFID2006AA04A1101.2RadioFrequencyRFID[1]RFIDForresterResearchRFID2004ITRFIDIBMTIPhilipSAPIT[2]RFIDRFIDRFIDRFIDRFRFIDRFIDRFID50KHz5.8GHzRFIDRFIDRFIDRFRFRFRFTransceiverRFIDRFRFIDReaderReader2Reader1.11.1RFIDRFID2003500RFIDRFID30830081.3IT1314RFIDRFID40RFID51012RFIDRFIDTorayDataconSUSSRFIDUnitPerHourUPH50008000UPH1996TAL4000TAL80004000UPH38000UPH,2004TAL5000TAL100005000UPH10000UPHTAL40001008800100RFID1.31.3.11958ICICICICICBACK-ENDMCMCSPBGAICPCBCOBCOFFCOBFCOF1.31.3IC12080PTHSIPDIPPGAPTH42080SMTPTHPCBSOPPLCCPQFPQFJLCCC2090BGAFCTWLPMCM20SiPSiPSiP[3,4]WireBondingTABFlipChipBondingVLSIULSII/O50mµ2020mmmm×1200/I/O100msTABTABICICIC[3-5]CSPWLP1.3.25[6]3-D123456PC1.21.2PC1RS-170/CCIRY/CRGB2LED34PC/PCIPCI64CompactPCIPC104ISA75PCPCCPU6PASS/FAILC/C++ActiveXLCDBGA/7I/OI/O//7PCRFID1.3.3ICICSchulzeMarkA.[7]8KubotaToshiro[8]micropipesGeodesicActiveContoursYehChi-hao[9]0805PriceJefferyR.[10]HouTung-HsuTony[11]SOPDIPYeQ.Z.[12]ICCCDICSMTKoshimizuHiroyasu[13]KassimA.A.[14]ICnotchdimpleHildebrandLars[15]HSI1.3.4,[16]9CCDCCDCCDCCDCCDCCDCCD[17]BartschatCCD[18]CCDCCDCCDCCDSMTSMT2505[17]CCDCCDKarlSussFC250[19]TorayFC2000[20]KarlSussFC-250CCD[19]10aHudson1998bHudson20011.4Tews[21]Hudson1998[22]Hudson19981.4aCCDCCDHudson2001[23]Hudson2001CCD1.4bCCDLink2003CCDCCDCCDCCD3RFID1.4RFID11RFIDRFIDXY20µm1RFID122.2.1[24-26]RFID2.2[27](),fxy(),Fuv(),Guv(),gxySMTCCDLee13Turky19711x2x…nxn12...iiinxxx≤≤≤}{1212,,...mnnixMedxxxx+==2.1[28][29]TSHuangmn2mnm×−×1)/2htmn=2)HedMedMedLM3)gPP[][]1ggHPHP=−4)gPP[][]1ggHPHP=+gedPM1ededLMLM=+5)edhLMt6[]edededLMLMHM=+1ededMM=+edhLMt≥76)1ededMM=−[]edededLMLMHM=−edhLMt≤7)38)22.32.3.114(,)fxyT255(,)(,)0(,)fxyTFxyfxyT=≤2.231)[30,31]T2.1Tab2.12)153)lZHZ0K=()KT()2KlHZZT+=2.3(0)TOZBZ()()(,)(,)(,)(,)(,)KKZijTOZijTZijNijZNij⋅=∑∑2.4()()(,)(,)(,)(,)(,)KKZijTBZijTZijNijZNij⋅=∑∑2.5(,)Zij(,)ij(,)Nij(,)ij(,)1.0Nij=(1)2KOBZZT++=2.6(1)()KKTT+=()KT1KK=+(1)()KKTT+=1()px2()px162111111()exp22xpxµσπσ−=−2.72222211()exp22xpxµσπσ−=−2.81µ2µ1σ2σ1122()()()pxPpxPpx=[phone]expexp2222PxPxµµσσπσπσ−−=−+−2.91µ2µ121PP+=12µµT()1ET()2ET()()11TETpxdx−∞=∫()()22TETpxdx∞=∫2.10()()()2211TTETPpxdxPpxdx∞−∞=+∫∫2.11()0ETT∂=∂2.11()()()2222222222212121221121221212ln2220PTPσσσσσσµσµσµσµσ+−+−+−=2.122212σσσ==2122121ln2PTPµµσµµ+=+−2.131µ2µ2σ1P2P17Otsu1978[32]2wσ2Bσ2Tσ322Bwσλσ=22Twσκσ=22BTσησ=2.14222()()()AaBbTwwσµµµµ=−+−2.152σAwAaµABwBbµBµTAB2()TσT2.3.2[33]XYmaxXminXmaxYminY18maxminmaxminminminYYXXYYXX−−=−−2.16XmaxminmaxminmaxminmaxminmaxminYYXYYXYXXXXX−−=+−−2.17maxminmaxminYYaXX−=−maxminmaxminmaxminXYYXbXX−=−2.18YaXb=+2.19abXY1amaxminmaxminYYXX−−bb1)maxYminYb2)