Domov technika Machine vision

Machine vision



Introduction

Machinevisionisacomprehensivetechnology,includingimageprocessing,mechanicalengineeringtechnology,control,electriclightsourcelighting,opticalimaging,sensors,analoganddigitalvideotechnology,computersoftwareandhardwareTechnology(imageenhancementandanalysisalgorithms,imagecards,I/Ocards,etc.).Atypicalmachinevisionapplicationsystemincludesimagecapture,lightsourcesystem,imagedigitizationmodule,digitalimageprocessingmodule,intelligentjudgmentdecisionmoduleandmechanicalcontrolexecutionmodule.

Themostbasicfeatureofamachinevisionsystemistoimprovetheflexibilityandautomationofproduction.Insomedangerousworkingenvironmentsthatarenotsuitableformanualoperationsoroccasionswhereartificialvisionisdifficulttomeettherequirements,machinevisionisoftenusedtoreplaceartificialvision.Atthesametime,inthemassrepetitiveindustrialproductionprocess,theuseofmachinevisioninspectionmethodscangreatlyimprovetheefficiencyandautomationofproduction.

Development

Nowadays,Chinaisbecomingoneofthemostactiveregionsinthedevelopmentofmachinevisionintheworld,withapplicationscoveringindustry,agriculture,medicine,military,aerospace,meteorology,astronomy,andpublicsecurity,Transportation,security,scientificresearchandotherindustriesinthenationaleconomy.TheimportantreasonisthatChinahasbecometheprocessingcenteroftheglobalmanufacturingindustry.Thehigh-demandpartsprocessingandthecorrespondingadvancedproductionlineshaveenabledmanyinternationallyadvancedmachinevisionsystemsandapplicationexperiencetoenterChina.

Afteralongperiodofdormancy,China'smachinevisionmarketusheredinexplosivegrowthin2010.ThedatashowsthatChina'smachinevisionmarketreached830millionyuanthatyear,anincreaseof48.2%year-on-year.Amongthem,thegrowthrateofsmartcameras,software,lightsourcesandboardsallreached50%,andindustrialcamerasandlensesalsomaintainedagrowthrateofmorethan40%.,Arethehighestlevelsince2007.

In2011,China'smachinevisionmarketenteredapost-growthadjustmentperiod.Comparedwiththerapidgrowthin2010,althoughthegrowthratehasdeclined,itstillmaintainsahighlevel.In2011,thescaleofChina’smachinevisionmarketwas1.08billionyuan,ayear-on-yearincreaseof30.1%,andthegrowthratedropped18.1percentagepointsfrom2010.Amongthem,smartcameras,industrialcameras,softwareandboardsallmaintainedagrowthrateofnolessthan30%.Thelightsourcehasalsoreachedagrowthrateof28.6%,whichismuchhigherthanthegrowthrateofChina'soverallautomationmarket.Theelectronicsmanufacturingindustryisstillthemainfactordrivingtherapidgrowthofdemand.In2011,themarketsizeoftheelectronicsmanufacturingindustryformachinevisionproductswasRMB500million,anincreaseof35.1%.Themarketsharereached46.3%.Electronicmanufacturing,automobiles,pharmaceuticalsandpackagingmachineryaccountfornearly70%ofthemachinevisionmarketshare.

Overview

Machinevision

Thecharacteristicofthemachinevisionsystemistoimprovetheflexibilityandautomationofproduction.Insomedangerousworkingenvironmentsthatarenotsuitableformanualoperationsoroccasionswhereartificialvisionisdifficulttomeettherequirements,machinevisionisoftenusedtoreplaceartificialvision;atthesametime,inthemassindustrialproductionprocess,manualvisualinspectionofproductqualityisinefficientandinaccurate.,Theuseofmachinevisioninspectionmethodscangreatlyimproveproductionefficiencyandthedegreeofautomationofproduction.Moreover,machinevisioniseasytorealizeinformationintegration,anditisthebasictechnologytorealizecomputerintegratedmanufacturing.

Basicstructure

Atypicalindustrialmachinevisionsystemincludes:lightsource,lens(fixedfocuslens,zoomlens,telecentriclens,microscopelens),camera(includingCCDcamera)AndCOMScamera),imageprocessingunit(orimagecapturecard),imageprocessingsoftware,monitor,communication/inputoutputunit,etc.

Thesystemcanbefurtherdividedinto

First,thecollectionandanalysissystemareseparated.

HostComputer

FrameGrabberandimageprocessor

Videocamera

Fixfocuslenslens

Microscopelens

Lightingequipment

HalogenlightsourceLEDlightsource

Highfrequencyfluorescentlightsource

p>

Flashsource

Otherspeciallightsources

Imagedisplay

LCD

Mechanismandcontrolsystem

PLC,PC-Basecontroller

Precisiontable

Servomotionmachine

Two,collectionandanalysisintegratedsystem

b>

Smartcamera(integratedimageacquisitionandanalysis)

Othersupportingperipheralequipment:lightsource,display,PLCcontrolsystem,etc.

Workingprinciple

ThemachinevisioninspectionsystemusesaCCDcameratoconvertthedetectedtargetintoanimagesignal,whichissenttoadedicatedimageprocessingsystem,accordingtothepixeldistributionandbrightness,colorandotherinformation,Convertedintodigitalsignals,theimageprocessingsystemperformsvariousoperationsonthesesignalstoextractthecharacteristicsofthetarget,suchasarea,number,position,length,andthenoutputtheresultsaccordingtothepresetallowabilityandotherconditions,includingsize,angle,andsize.Count,pass/fail,yes/no,etc.,torealizetheautomaticidentificationfunction.

Typicalstructure

Atypicalmachinevisionsystemincludesthefollowingfiveblocks:

Lighting

LightingaffectstheinputofthemachinevisionsystemImportantfactors,itdirectlyaffectsthequalityoftheinputdataandtheapplicationeffect.Sincethereisnogeneralmachinevisionlightingequipment,foreachspecificapplicationexample,thecorrespondinglightingdeviceshouldbeselectedtoachievethebesteffect.Thelightsourcecanbedividedintovisiblelightandinvisiblelight.Severalcommonlyusedvisiblelightsourcesareincandescentlamps,fluorescentlamps,mercurylampsandsodiumlamps.Thedisadvantageofvisiblelightisthatthelightenergycannotbekeptstable.Howtokeeplightenergystabletoacertainextentisanurgentproblemthatneedstobesolvedintheprocessofpracticalapplication.Ontheotherhand,ambientlightmayaffectthequalityoftheimage,soaprotectivescreencanbeusedtoreducetheimpactofambientlight.Thelightingsystemcanbedividedintobacklighting,forwardlighting,structuredlightandstroboscopiclightingaccordingtoitsilluminationmethod.Amongthem,thebacklightingisthattheobjecttobemeasuredisplacedbetweenthelightsourceandthecamera,anditsadvantageisthatitcanobtainhigh-contrastimages.Theforwardilluminationisthatthelightsourceandthecameraarelocatedonthesamesideoftheobjectundertest,whichisconvenientforinstallation.Structuredlightilluminationistoprojectagratingoralinelightsourceontothemeasuredobject,anddemodulatethethree-dimensionalinformationofthemeasuredobjectaccordingtothedistortionproducedbythem.Stroboscopiclightingistoirradiatehigh-frequencylightpulsesontheobject,andthecamerashootingrequiressynchronizationwiththelightsource.

Lens

FOV(FieldofVision)=requiredresolution*sub-pixel*camerasize/PRTM(partmeasurementtoleranceratio)

lensselectionNote:

①Focallength②Targetheight③Imageheight④Magnification⑤distancefromimagetotarget⑥centerpoint/node⑦distortion

Howtodeterminethefocallengthofthelensinvisualinspection

Whenchoosingasuitableindustriallensforaspecificapplication,thefollowingfactorsmustbeconsidered:

·Fieldofview-thesizeoftheimagedarea.

·Workingdistance(WD)-thedistancebetweenthecameralensandtheobservedobjectorarea.

·CCD-thesizeofthecameraimagingsensordevice.

·Thesefactorsmustbetreatedinaconsistentmanner.Ifyouaremeasuringthewidthofanobject,youneedtousethehorizontalCCDspecifications,andsoon.Ifthemeasurementismadeininches,thecalculationismadeinfeet,andfinallyconvertedtomillimeters.

Refertothefollowingexample:Thereisa1/3”C-mountCCDcamera(4.8mmhorizontally).Thedistancebetweentheobjectandthefrontofthelensis12”(305mm).Thesizeofthefieldofvieworobjectis2.5”(64mm).Theconversionfactoris1”=25.4mm(rounded).

FL=4.8mmx305mm/64mm

FL=1464mm/64mm

FL=Accordingtotherequirementsof23mmlens

FL=0.19”x12”/2.5”

FL=2.28”/2.5”

FL=0.912”x25.4mm/inch

FL=Accordingtotherequirementsofthe23mmlens

Note:Donotconfusetheworkingdistancewiththedistancefromtheobjecttotheimage.Theworkingdistanceisthedistancefromthefrontoftheindustriallenstotheobjectbeingobserved.ThedistancefromtheobjecttotheimageisthedistancebetweentheCCDsensorandtheobject.Whencalculatingtherequiredfocallengthoftheindustriallens,theworkingdistancemustbeused.

High-speedcamera

Camera(2photos)

Accordingtodifferentstandards,itcanbedividedinto:Standardresolutiondigitalcamerasandanalogcameras,etc.Differentcamerasandhigh-resolutioncamerasshouldbeselectedaccordingtodifferentpracticalapplications:

Dividedbyimagingcolor,itcanbedividedintocolorcamerasandblackandwhitecameras;

Dividedbyresolution,thenumberofpixelsis38Thosewithlessthan10,000areordinarytype,andthehigh-resolutiontypewithpixelsabove380,000;

Dividedaccordingtothesizeofthephotosensitivesurface,itcanbedividedinto1/4,1/3,1/2,1inchCamera;

Accordingtothescanningmethod,itcanbedividedintotwomethods:linescancamera(linescancamera)andareascancamera(areascancamera);(areascancameracanbedividedintointerlacedscancameraandProgressivescancamera);

Accordingtothesynchronizationmethod,itcanbedividedintoordinarycameras(internalsynchronization)andcameraswithexternalsynchronizationfunctions.

Framegrabber

Acquisitioncard(2photos)

Framegrabberisonlyapartofacompletemachinevisionsystem,butitplaysAveryimportantrole.Theframegrabberdirectlydeterminestheinterfaceofthecamera:blackandwhite,color,analog,digitalandsoon.

APCIorAGPcompatiblecapturecardismoretypical,whichcanquicklytransferimagestocomputermemoryforprocessing.Somecapturecardshavebuilt-inmultiplexers.Forexample,youcanconnect8differentcameras,andthentellthecapturecardtousetheinformationcapturedbywhichcamera.Somecapturecardshaveabuilt-indigitalinputtotriggerthecapturecardtocapture.Whenthecapturecardcapturesanimage,thedigitaloutputporttriggersthegate.

Visionprocessor

Thevisionprocessorintegratesacapturecardandaprocessor.Inthepast,whenthecomputerspeedwasslow,thevisualprocessorwasusedtospeedupthevisualprocessingtask.Thecapturecardtransmitstheimagetothememory,andthencalculatesandanalyzes.ThecurrentmainstreamconfigurationofPLC,andtheconfigurationishigher,thevisionprocessorhasalmostwithdrawnfromthemarket.

Machineselection

Inamachinevisionsystem,itisveryimportanttoobtainahigh-qualityprocessableimage.Thereasonforthesuccessofthesystemistoensurethattheimagequalityisgoodandthefeaturesareobvious.Thefailureofamachinevisionprojectismostlyduetopoorimagequalityandunobviousfeatures.Toensureagoodimage,asuitablelightsourcemustbeselected.

Basicelementsoflightsourceselection:

Contrast:Contrastisveryimportantformachinevision.Themostimportanttaskoflightingformachinevisionapplicationsistomaximizethecontrastbetweenthefeaturesthatneedtobeobservedandtheimagefeaturesthatneedtobeignored,sothatthefeaturescanbeeasilydistinguished.Contrastisdefinedasasufficientamountofgrayscaledifferencebetweenafeatureanditssurroundingarea.Goodlightingshouldbeabletoensurethatthefeaturestobedetectedstandoutfromotherbackgrounds.

Brightness:Whenchoosingtwolightsources,thebestchoiceistochoosethebrighterone.Whenthelightsourceisnotbrightenough,theremaybethreebadsituations.First,thesignal-to-noiseratioofthecameraisnotenough;becausethebrightnessofthelightsourceisnotenough,thecontrastoftheimageisboundtobeinsufficient,andthepossibilityofnoiseintheimageincreasesimmediately.Secondly,thebrightnessofthelightsourceisnotenough,andtheaperturemustbeincreased,therebyreducingthedepthoffield.Inaddition,whenthebrightnessofthelightsourceisinsufficient,randomlightsuchasnaturallightwillhavethegreatestimpactonthesystem.

Robustness:Anotherwaytotestagoodlightsourceistoseeifthelightsourceisleastsensitivetothepositionofthepart.Whenthelightsourceisplacedindifferentareasordifferentanglesofthecamera'sfieldofview,theresultingimageshouldnotchangeaccordingly.Thelightsourcewithstrongdirectivityincreasesthepossibilityofspecularreflectioninthehighlightarea,whichisnotconducivetothesubsequentfeatureextraction.

Agoodlightsourceneedstobeabletomakethefeaturesyouneedtolookforveryobvious.Inadditiontothecamerathatcancapturetheparts,agoodlightsourceshouldbeabletoproducemaximumcontrast,sufficientbrightnessandinsensitivetochangesinthepositionoftheparts.Oncethelightsourceisselected,therestoftheworkismucheasier.Thespecificlightsourceselectionmethodalsoliesinthepracticalexperienceoftheexperiment.

Applicationcase

Intheclothproductionprocess,highlyrepetitiveandintelligenttaskslikeclothqualityinspectioncanonlybedonebymanualinspection,behindthemodernassemblylineItcanoftenbeseenthatmanyinspectionworkersperformthisprocess,whichaddshugelaborandmanagementcoststothecompany,butitstillcannotguaranteea100%inspectionpassrate(ie"zerodefect").Theinspectionofclothqualityisrepetitivework,error-proneandlowefficiency.

Automatictransformationoftheassemblylinemakestheclothproductionassemblylineafast,real-time,accurateandefficientassemblyline.Ontheassemblyline,thecolorandquantityofallclothsmustbeautomaticallyconfirmed(hereinafterreferredtoas"clothinspection").Theautomaticrecognitiontechnologyofmachinevisionisusedtocompletetheworkpreviouslydonebyhumans.Inmassinspectionofcloth,manualinspectionofproductqualityisinefficientandinaccurate.Usingmachinevisioninspectionmethodscangreatlyimproveproductionefficiencyandautomation.

Featureextractionandidentification

Generalclothdetection(automaticidentification)firstuseshigh-definition,high-speedcameralenstocapturestandardimages,andthensetsAcertainstandard;thentakethedetectedimageandcomparethetwo.Butitismorecomplicatedintheclothqualityinspectionproject:

1.Thecontentoftheimageisnotasingleimage,andthenumber,size,color,andlocationofimpuritiesineachareatobetestedmaynotbethesame.

2.Theshapeofimpuritiesisdifficulttodetermineinadvance.

3.Becausetherapidmovementoftheclothreflectslight,theremaybealotofnoiseintheimage.

4.Therearereal-timerequirementsforfabricinspectionontheassemblyline.

Duetotheabovereasons,correspondingalgorithmsshouldbeadoptedinimagerecognitionprocessingtoextractthecharacteristicsofimpurities,performpatternrecognition,andrealizeintelligentanalysis.

Colordetection

Generallyspeaking,theimagesobtainedfromcolorCCDcamerasareallRGBimages.Thatistosay,eachpixeliscomposedofthreecomponents,red(R),green(G),andblue(B),torepresentapointintheRGBcolorspace.Theproblemisthatthesechromaticaberrationsaredifferentfromtheperceptionofthehumaneye.Evenasmallamountofnoisecanchangethepositioninthecolorspace.Sonomatterhowsimilarourhumaneyesfeel,theyarenotallthesameinthecolorspace.Basedontheabovereasons,weneedtoconvertRGBpixelsintoanothercolorspaceCIELAB.Thepurposeistomakeourhumaneyesfeelascloseaspossibletothecolordifferenceinthecolorspace.

Blobdetection

Accordingtotheprocessedimageobtainedabove,accordingtotherequirements,detecttheimpuritystainsunderapurecolorbackground,andcalculatetheareaof​​theoutstandingspots,Todeterminewhetheritiswithinthedetectionrange.Therefore,theimageprocessingsoftwaremusthavethefunctionofseparatingthetarget,detectingthetarget,andcalculatingitsarea.

Blobanalysis(BlobAnalysis)istoanalyzetheconnecteddomainofthesamepixelintheimage,andtheconnecteddomainiscalledBlob.ThestainsintheimageprocessedbyBinaryThresholdingcanbeconsideredasblobs.TheBlobanalysistoolcanseparatethetargetfromthebackground,andcancalculatethenumber,position,shape,directionandsizeofthetarget,andcanalsoprovidethetopologicalstructurebetweentherelevantspots.Intheprocessofprocessing,insteadofanalyzingindividualpixelsonebyone,itoperatesonlinesofgraphics.Eachlineoftheimageusesrun-lengthencoding(RLE)torepresenttheadjacenttargetrange.Comparedwiththepixel-basedalgorithm,thisalgorithmgreatlyimprovestheprocessingspeed.

Resultprocessingandcontrol

Theapplicationsavesthereturnedresultsintothedatabaseorthelocationspecifiedbytheuser,andcontrolsthemechanicalparttoperformcorrespondingmovementsaccordingtotheresults.

Accordingtotheidentificationresults,itisstoredinthedatabaseforinformationmanagement.Inthefuture,theinformationcanbesearchedandinquiredatanytime,andthemanagercanknowtheavailabilityoftheassemblylineduringacertainperiodoftime,makearrangementsforthenextwork;canknowthequalityoftheinnercloth,andsoon.

Applicationstatus

Inforeigncountries,theapplicationofmachinevisionismainlyreflectedinthesemiconductorandelectronicindustries,ofwhichabout40%-50%areconcentratedinsemiconductorsindustry.Specifically,suchasPCBprintedcircuit:variousproductionprintedcircuitboardassemblytechnologiesandequipment;single,double-sided,multi-layercircuitboards,coppercladlaminatesandrequiredmaterialsandauxiliarymaterials;auxiliaryfacilitiesandconsumables,inks,medicines,accessories,andelectronics;Packagingtechnologyandequipment;screenprintingequipmentandscreenperipheralmaterials,etc.SMTsurfacemount:SMTprocessandequipment,solderingequipment,testequipment,reworkequipmentandvariousauxiliarytoolsandaccessories,SMTmaterials,patchtablets,adhesives,flux,solderandanti-oxidationoil,solderpaste,cleaningagents,etc.;Reflowsolderingmachine,wavesolderingmachineandautomatedproductionlineequipment.Electronicproductionandprocessingequipment:electroniccomponentmanufacturingequipment,semiconductorandintegratedcircuitmanufacturingequipment,componentmoldingequipment,electronictoolsandmolds.Themachinevisionsystemhasalsobeenwidelyusedinallaspectsofqualityinspection,anditsproductsoccupyapivotalpositionintheapplication.Inaddition,machinevisionisalsousedinvariousotherfields.

InChina,theapplicationofvisiontechnologybeganinthe1990s,becausetheindustryitselfisanemergingfield,coupledwiththeinsufficientpopularityofmachinevisionproducttechnology,resultinginalmostblankapplicationsintheaboveindustries.Atpresent,mostdomesticmachinevisionsareforeignbrands.Mostdomesticmachinevisioncompaniesbasicallystartedbyactingasagentsforvariousforeignmachinevisionbrands.Withthecontinuousapplicationofmachinevision,thecompanyhasgraduallygrowninscaleandhasgraduallymaturedintechnology.

Withtheimprovementoftheeconomiclevel,3Dmachinevisionhasalsobeguntoenterpeople'sfieldofvision.3Dmachinevisionismostlyusedfortheratingoffruitsandvegetables,wood,cosmetics,bakedgoods,electroniccomponentsandpharmaceuticalproducts.Itcanimprovetheproductioncapacityofqualifiedproductsandscrapinferiorproductsearlyintheproductionprocess,therebyreducingwasteandsavingcosts.Thisfunctionisverysuitableforimagingofproductattributessuchasheight,shape,quantityandevencolor.

Intermsofindustryapplications,therearemainlypharmaceutical,packaging,electronics,automobilemanufacturing,semiconductor,textile,tobacco,transportation,logisticsandotherindustries.Usingmachinevisiontechnologytoreplacelaborcanimproveproductionefficiencyandproductquality.Forexample,inthelogisticsindustry,machinevisiontechnologycanbeusedforsortingandsortingofexpressdelivery,andmostexpresscompanieswillnotmanuallysort,reducethedamagerateofitems,improvesortingefficiency,andreducemanuallabor.

Generationanddevelopment

TheresearchofmachinevisionstartedfromtheresearchoftheAmericanscholarL.R.Robertsonunderstandingthebuildingblockworldcomposedofpolyhedronsinthemid-1960s.Thetechnologiesusedatthattime,suchaspreprocessing,edgedetection,contourformation,objectmodeling,andmatching,havebeenappliedinmachinevisioneversince.Robertsadoptedabottom-upapproachintheimageanalysisprocess.Edgedetectiontechnologyisusedtodeterminecontourlines,andareaanalysistechnologyisusedtodividetheimageintoareascomposedofpixelswithsimilargraylevels.Thesetechnologiesarecollectivelycalledimagesegmentation.Itspurposeistodescribetheanalyzedimagewithcontourlinesandregions,soastocompareandmatchwiththemodelstoredinthemachine.Practicehasshownthatitistoodifficulttouseonlybottom-upanalysis,andtop-downanalysismethodsmustbeusedatthesametime,thatis,thetargetisdividedintoseveralsub-targets,andheuristicknowledgeisusedtopredictthetarget.Thisisconsistentwiththebottom-upandtop-downmethodsusedinspeechcomprehension.Intheresearchofimageunderstanding,A.Guzmanproposedtheuseofheuristicknowledge,showingthatthemethodofexplainingcontourpaintingbysymbolicprocessdoesnothavetoresorttonumericalcalculationprogramssuchasleastsquaresmatching.

Inthe1970s,machinevisionformedseveralimportantresearchbranches:①target-guidedimageprocessing;②parallelimageprocessingandanalysisalgorithms;③extractingthree-dimensionalinformationfromtwo-dimensionalimages;④sequentialimageanalysisandmotionParameterevaluation;⑤Expressionofvisualknowledge;⑥Knowledgebaseofvisionsystem,etc.

Thelatestdiscovery

TheAchilles’heelofmachinevision:AccordingtoMIT’s“TechnologyReview”,researchersfromGoogleandtheOpenAIInstitutehavediscoveredmachinevisionalgorithmsOneoftheweaknesses:machinevisionwillbedisturbedbysomemodifiedimages,andhumanscaneasilyfindthemodificationoftheseimages.

Applicationfield

Theapplicationofmachinevisionmainlyincludesinspectionandrobotvision:

⒈Inspection:Itcanbedividedintohigh-precisionquantitativeinspection(forexample,Cellclassificationofphotomicrographs,sizeandpositionmeasurementofmechanicalparts)andqualitativeorsemi-quantitativetestingwithoutmeasuringinstruments(suchasproductappearanceinspection,identificationandpositioningofpartsontheassemblyline,defectinspectionandassemblycompletenessinspection).[Thatis,thehopperpickingproblem).Asforoperationsandactionsinasmallarea,tactilesensingtechnologyisalsoneeded.

Inaddition,thereare:

1Automaticopticalinspection

2Facerecognition

3Self-drivingcars

4Productqualityclassification

5Automaticdetectionofprintingquality

6Textrecognition

7Texturerecognition

8Trackingandpositioning

......

andothermachinevisionimagerecognitionapplications.

[MachineVisionFeatures]

⒈Thecamera'sphotographingspeedisautomaticallymatchedwiththespeedofthemeasuredobject,andtheidealimageistaken;

⒉Thesizeofthepartsrangesfrom2.4mmto12mm,andthethicknesscanbedifferent;

⒊Thesystemselectsworkpiecesofdifferentsizesaccordingtotheoperator,callsthecorrespondingvisionprogramforsizedetection,andoutputstheresults;

⒋Forpartsofdifferentsizes,thesortingdeviceandtheconveyingdevicecanpreciselyadjustthewidthofthematerialpath,sothatthepartscanmoveonafixedpathandperformvisualinspection;

⒌MachinevisionsystemdistinguishesTheratereaches2448×2048,andthedynamicdetectionaccuracycanreach0.02mm;

⒍Themisseddetectionrateofwasteproductsis0;

⒎Thesystemcanmonitorthedetectionprocessbydisplayingimages,anditcanalsopassTheinspectiondatadisplayedontheinterfacecandynamicallyviewtheinspectionresults;

⒏Ithasthefunctionoftimelyandaccuraterejectioncontrolsignalsandrejectsthewrongparts;

⒐Thesystemcanself-checkitsmainequipmentWhetherthestateisnormalornot,equippedwithastatusindicator;atthesametime,differentoperatingpermissionscanbesetforsystemmaintenancepersonnelandusers;

⒑Real-timedisplayofthedetectionscreen,Chineseinterface,youcanbrowseseveralunqualifiedimages,Ithasthefunctionofstoringandviewingtheimageofthewrongworkpieceinrealtime;

⒒cangeneratetheerrorresultinformationfile,includingthecorrespondingerrorimage,andprintitout.

Applicationexamples

⒈Intelligentintegratedtestsystemforinstrumentpanelassemblybasedonmachinevision

EQ140-ⅡautomobileinstrumentpanelassemblyisproducedbyaChineseautomobilecompanyForinstrumentproducts,theinstrumentpanelisequippedwithspeedodometer,watertemperaturemeter,gasolinemeter,ammeter,signalwarninglight,etc.Theproductionbatchislarge,andafinalqualityinspectionisrequiredbeforeleavingthefactory.Testingitemsinclude:testingtheindicationerrorsoffivemeterpointerssuchasthespeedometer;testingwhether24signalwarninglightsand9lightinglightsaredamagedormissing.Generally,manualvisualinspectionisusedforinspection,whichhaslargeerrorsandpoorreliability,whichcannotmeettheneedsofautomatedproduction.Theintelligentintegratedtestsystembasedonmachinevisionhaschangedthisstatusquo,realizedtheintelligent,fullyautomatic,high-precision,andfastqualityinspectionofthedashboardassembly,overcomevariouserrorscausedbymanualinspection,andgreatlyimprovedtheinspection.efficient.

Thewholesystemisdividedintofourparts:integratedmulti-channelstandardsignalsourcethatprovidesanalogsignalsourcefordashboard,dual-coordinateCNCsystemwithimageinformationfeedbackpositioning,cameraimageacquisitionsystemandmaster-slavemachineParallelprocessingsystem.

⒉Automaticdamagecontrolsystemforthesurfaceofmetalplates

Thesurfacequalityofmetalplates,suchaslargepowertransformercoils,flatwireradios,andhazyskins,havehighrequirements,buttheoriginaluseofartificialTheinspectionmethodofvisualinspectionorusingadialindicatorplusacontrolneedleisnotonlysusceptibletosubjectivefactors,butmayalsocausenewscratchesonthesurfacetobemeasured.Theautomaticflawdetectionsystemforthemetalplatesurfaceusesmachinevisiontechnologytoautomaticallyinspectthemetalsurfacedefects,andthedetectioniscarriedoutathighspeedandaccuracyduringtheproductionprocess.Atthesametime,duetothenon-connectedanglemeasurement,thepossibilityofnewscratchesisavoided.TheworkingprinciplediagramisshowninFigure8-6;inthissystem,alaserisusedasthelightsource,thestraylightaroundthelaserbeamisfilteredthroughapinholefilter,andthebeamexpanderandcollimatormakethelaserbeamintoparallellightAnduniformlyilluminatethesurfaceofthemetalplatetobeinspectedwithanincidentangleof45degrees.Themetalplateisplacedontheinspectiontable.TheinspectiontablecanbemovedinthethreedirectionsofX,YandZ.ThecameraadoptsTCD142Dtype2048lineChenCCD,andthelensadoptsordinarycameralens.TheCCDinterfacecircuitadoptsasingle-chipmicrocomputersystem.ThehostPCmainlycompletesimagepreprocessinganddefectclassificationordepthcalculationofscratches,etc.,andcandisplaythedetecteddefectsorscratchimagesonthedisplay.Two-waycommunicationbetweentheCCDinterfacecircuitandthePCthroughtheRS-232port,combinedwiththeasynchronousA/Dconversionmethod,constitutesahuman-machineinteractivedataacquisitionandprocessing.

Thissystemmainlyusesthecombinationoftheself-scanningcharacteristicsofthelinearCCDandtheX-directionmovementofthesteelplatetobeinspectedtoobtainthree-dimensionalimageinformationonthesurfaceofthemetalplate.

⒊Automobilebodyinspectionsystem

The100%on-lineinspectionofthecontourdimensionalaccuracyofthe800seriesoftheBritishROVERautomobilecompanyisatypicalexampleofthemachinevisionsystemusedinindustrialinspection.Forexample,thesystemconsistsof62measurementunits,eachmeasurementunitincludesalaserandaCCDcameratodetect288measurementpointsonthecarbodyshell.Thecarbodyisplacedunderthemeasuringframe,andtheprecisepositionofthecarbodyiscalibratedthroughsoftware.

Thecalibrationofthemeasuringunitwillaffectthedetectionaccuracy,soitispaidspecialattention.Eachlaser/cameraunitiscalibratedoffline.Atthesametime,thereisalsoacalibrationdevicethathasbeencalibratedwithathree-coordinatemeasuringmachineinanofflinestate,whichcanperformonlinecalibrationofthecameratop.

Thedetectionsystemdetectsthespeedofacarbodyevery40seconds,anddetectsthreetypesofcarbodies.ThesystemcomparesthetestresultwiththequalifiedsizeofthepersonandextractedfromtheCADmodel,andthemeasurementaccuracyis±0.1mm.ROVER'squalityinspectorsusethissystemtodeterminethesizeconsistencyofkeyparts,suchastheoverallappearanceofthecarbody,doors,andglasswindows.PracticehasprovedthatthesystemissuccessfulandwillbeusedforbodyinspectionofothervehiclesinROVER'ssystem.

⒋Papercurrencyprintingqualityinspectionsystem:

Thissystemusesimageprocessingtechnologytopassmorethan20characteristics(number,braille,color,pattern,etc.)ofbanknotesonthebanknoteproductionlinePerformcomparativeanalysistodetectthequalityofbanknotesandreplacethetraditionalhumaneyediscriminationmethod.

⒌Intelligenttrafficmanagementsystem:

Byplacingacameraonthetrafficarteries,whenthereareillegalvehicles(suchasrunningaredlight),thecamerawilltakepicturesofthelicenseplateofthevehicleandtransmitittothecentralmanagementThesystemusesimageprocessingtechnologytoanalyzethepicturestaken,extractthelicenseplatenumber,andstoreitinthedatabaseformanagementpersonneltoretrieve.

⒍Metallographicanalysis:

Themetallographicimageanalysissystemcanaccuratelyandobjectivelyanalyzethematrixstructure,impuritycontent,andtissuecompositionofmetalsorothermaterials.Qualityprovidesareliablebasis.

⒎Medicalimageanalysis:

Automaticclassificationandcountingofbloodcells,chromosomeanalysis,cancercellidentification,etc.

⒏Bottledbeerproductionlinedetectionsystem:

Itcandetectwhetherthebeerreachesthestandardcapacityandwhetherthebeerlabeliscomplete

⒐Large-scaleworkpieceparallelismandperpendicularitymeasuringinstrument:

Thelarge-scaleworkpieceparallelismandperpendicularitymeasuringinstrumentadoptslaserscanningandCCDdetectionsystem.ItusesastablecollimatedlaserbeamasthemeasurementbaselineandisequippedwithRotatetheaxissystem,rotatethepentagramprismtoscanoutthereferenceplanesthatareparallelorperpendiculartoeachother,andcompareitwiththefacesofthelargeworkpiecetobetested.Whenprocessingorinstallinglargeworkpieces,theerrordetectorcanbeusedtomeasuretheparallelismandperpendicularitybetweenthesurfaces.

⒑Detectordeviceforthecontoursizeoftherebar:

Usestroboscopiclightastheilluminationsource,anduseareaarrayandlineararrayCCDasthedetectorforthecontoursizeoftherebartoachievethermalDynamicdetectionsystemforonlinemeasurementofgeometricparametersofrolledrebar.

⒒Bearingreal-timemonitoring:

Visualtechnologymonitorstheloadandtemperaturechangesofthebearinginrealtime,eliminatingthedangerofoverloadandoverheating.Thetraditionalpassivemeasurementofmeasuringthesurfaceoftheballtoensureprocessingqualityandsafeoperationbecomesactivemonitoring.

⒓Measurementofcracksonmetalsurface:

Usingmicrowaveasasignalsource,accordingtothemicrowavegeneratortosendoutsquarewaveswithdifferentwaverates,measurethecracksonthemetalsurfaceandthefrequencyofmicrowavewavesThehigherthecrack,thenarrowerthemeasurablecrack.

Developmenttrends

Machinevisionhasthefollowingdevelopmenttrends.

Pricescontinuetodrop

Atpresent,themachinevisiontechnologyinmycountryisnotyetmature,mainlyrelyingonimportingacompletesetofsystemsfromabroad,andthepriceisrelativelyexpensive.Withtheadvancementoftechnologyandthefiercemarketcompetition,Thepricedrophasbecomeaninevitabletrend,whichmeansthatmachinevisiontechnologywillgraduallybeaccepted.

Functionsaregraduallyincreasing

Therealizationofmorefunctionsmainlycomesfromtheenhancementofcomputingpower,higherresolutionsensors,fasterscanningrateandimprovementofsoftwarefunctions,PCprocessingWhilethespeedofthedevicehasbeensteadilyincreasing,itspricehasalsobeendeclining,whichhaspromotedtheemergenceoffasterbuses,whichinturnallowlargerimageswithmoredatatobetransmittedandprocessedatfasterspeeds.

Productminiaturization

Theproductminiaturizationtrendallowstheindustrytopackmorepartsinasmallerspace,whichmeansthatmachinevisionproductsbecomesmaller,soTheycanbeusedinthelimitedspaceprovidedbythefactory.Forexample,inindustrialaccessories,LEDhasbecomethedominantlightsource.Itssmallsizemakesiteasytodetermineimagingparameters.Theirdurabilityandstabilityareverysuitableforfactoryequipment.

Moreintegratedproducts

Thedevelopmentofsmartcamerasheraldsthetrendofincreasingintegratedproducts.Smartcamerasintegratetheprocessor,lens,lightsource,input/outputinasingleboxDevicesandEthernet,telephonesandPDAshavepromotedthedevelopmentoffasterandcheaperreducedinstructionsetcomputers(RISC),whichmadetheemergenceofsmartcamerasandembeddedprocessorspossible.Similarly,theadvancementoffieldprogrammablegatearray(FPGA)technologyhasaddedcomputingfunctionstosmartcameras,andembeddedprocessorsandhigh-performanceframecollectorsforPCs.SmartcamerascombineFPGA,DSP,andDSPthathandlemostcomputingtasks.Themicroprocessorwillbemoreintelligent.

Prospects

Becausethemachinevisionsystemcanquicklyobtainalargeamountofinformation,itiseasytoautomaticallyprocess,anditisalsoeasytointegratewithdesigninformationandprocessingcontrolinformation.Therefore,inthemodernautomatedproductionprocess,Peopleusemachinevisionsystemsextensivelyinfieldssuchasworkingconditionmonitoring,finishedproductinspectionandqualitycontrol.

Butmachinevisiontechnologyismorecomplicated,andthebiggestdifficultyisthatthehumanvisualmechanismisnotyetclear.Peoplecanuseintrospectiontodescribetheprocessofsolvingaproblem,andthensimulateitwithacomputer.Butalthougheverynormalpersonisa"visualexpert",itisimpossibletodescribehisownvisualprocessbyintrospection.Therefore,buildingamachinevisionsystemisaverydifficulttask.

Itcanbepredictedthatwiththematurityanddevelopmentofmachinevisiontechnologyitself,itwillbemoreandmorewidelyusedinmodernandfuturemanufacturingenterprises.

Tento článek je ze sítě, nereprezentuje pozici této stanice. Uveďte prosím původ dotisku
HORNÍ