基于X射線的輪轂缺陷檢測(cè)系統(tǒng)設(shè)計(jì)-英文文獻(xiàn)

APPLICATIONINNDTOF

COMPUTERDIGITALX-RAYREALTIMEIMAGESYSTEM

Abstract

Nondestructivetesting(NDT)isoneofthemostcommonlyusedscientificmeanstodetecttheinternaldefectswithoutdamagingthetestedobjects,thereforeitisfoundwidelyapplicationsintheindustriesofmechanical,petrochemical,aerospaceetc.RadiographicTesting(RT)byX-rayiscommonlyusedNDTmethodatpresent,itsdisadvantagesareunavoidableforitslongtestperiod,highcostandenvironmentpollution.Withthedevelopmentofcomputertechnology,anemergingNDTmethodwithdigitalX-rayrealtimeimage(DXRTI)isfoundincreasedapplicationinindustry.Thispaperpresentsthefundamentalsandapplicationsofcomputernumericalimagetechnique(CNIT)inNDTwithDXRTItechnique,andintroducedthenewdevelopedDXRTIsoftware‘XCX2000’anditsapplicationsandeconomicbenefitinwelddefectsdetectingforLPGCylinders.

Keywords

computer,software,RealTimeImage,digitalimagesprocessing,X-ray,NDT

1.Introduction

TheprincipleofNDTwithDXRTImaybebrieflyexpressedbythetwoconversions:Afterpenetrationofmetal,X-raysignalswascapturedandconvertedintothevisualizedimagesbytheImageIntensifier,whichiscalledphotoelectricconversion.ThevisualizedimageswasshotbyveryfineresolutionvideocameraandthenwasconvertedintodigitalimagebyImageAcquisitionCard,whichiscalledA/Dconversion.Thedigitalimagewasprocessedwithcomputeranddisplaytheproperties,dimensionandlocationoftheinsidedefectsonscreen.Finally,thedefectgradewasassessedaccordingtotherelevantcodesandstandards.ThequalityandeffectofDXRTIisacceptablygoodcomparedwithradiographicfilms,sothatthelatercanbereplacedbytheformermethod.Moreover,withitsmuchhigherefficiency,lowercost,betterinspectioneffectiveness,remotertransmissionandmuchmoreconvenience,DXRTIhasprosperousfutureinthefieldofNDT.

2.Systemrequirements

DXRTINDTsystemiscomposedbyX-raysource,imageintensifier,opticalchannel,videocamera,computer,ImageAcquisitionCard,imagedisplayandstorageequipment(seeFig.1).TheX-raysourceisdifferentfromtheconventionaloneinitsconstantpotential,smallfocusandforcedconvectioncoolingsystem.

2.1TheComputer’sRequirements

IntelMMX166CPUorfaster,32MBRAM,2.0GHarddiskfreespace,2DGraphicAccelerateCard(4Mvideomemory),17”SVGAMonitor,CD-RWriter.

OperatingSystem:MicrosoftWindows95/98.

Fig.1BlockDiagramforNDTwithDXRTI

2.2SystemManagement

Forconvenientoperations,twocomputerscanbeconnectedtothesystemtoformapeertopeernetwork,oneforimageacquisitionandtheotherforimageassessment,andeachcomputerworkswithidenticalpurpose.Theresultimagesarestoredintheharddiskforassessmentandtransmittedforverylongdistanceviainternet.

3.Digitalimageprocessingbycomputer

Theprocessingofdigitalimagesbycomputerincludesthreeparts:imageacquisition,imagestreatmentandimagedisplay。

3.1ImageAcquisition

X-rayfromradiographicmachinepenetratesthetestobjectandconvertedintothevisibleimageonthescreenbytheimageintensifier.ThentheimagesarechangedintovideosignalsbythevideocameraandsenttotheImageAcquisitionCardfordigitalconversion.Finallythedigitalimagesaretobeprocessedbythecomputer.

Therearetwoimportantparametersintheimageacquisition:grayscaleand

resolutionoftheimages.Commonly,theA/Dconverteroftheimageacquisitionis8bitswithcolorsof256grayscales.Higherresolutionoftheimageacquisitionwilleffectivelyimprovethedefectsresolvingpower.Theresolutionoftheimageacquisitionshouldbenolessthan768X576dotlines.

3.2ImageProcessing

Themainpurposeofimageprocessingistoextractspecificinformationfromtheimagesforcomprehensivecomputeranalysisandtheidentificationofdifferentimages.

ImageSuperpositionfornoisereduction

Itisnoticedthatrandomnoiseinformationisunavoidableinimageacquisition,whichisoneoftheimportantfactorsaffectingthequalityoftheimages.Forthecaseofstableimageacquisition,oneoftheeffectiveandpracticalwaystoeliminatetherandomnoiseistosuperposetheinputimagescontinuously.Theoretically,thetimedependentnoisecanberemovedorfilteredcompletelyifsufficientframesoftheimagessuperposed.Therefore,framesuperpositioniscommonlyusedintheprocessingofimages.Thismethodisexpressedbythefollowingequation:

〔1〕

Where,g(x,y)standsforthedotgrayscaleoftheimage,Misthetotalnumbersofthesuperposedimageframes,fk(x,y)isthegrayscalevaluefortheKthframeimage.ThesamplingtestshowsthattheresultswillbesatisfactorywhenM≥16.

GrayScaleExpand

〔2〕Grayscaleconversiontechniqueiscommonlyused.Withthistechnique,theimagequalitymaybeimprovedbyadjustingthegrayscaleoftheimagefollowingsomespecifiedrules.Formostofthecases,theimagegrayscalesvarywithinverynarrowregimes,resultingintotheimagewithverypoorcontrastandvaguemargins.Bylinearconversion,thegrayscalesoftheimagecanbeexpandedto256(0～255),sothatthecontrastandtheresolvingpoweroftheimagewillbeimproved.SupposetheoriginalgrayscaleregimesofaimageisA1～A2,therelevantconversionequationis

〔3〕

Moreover,withpropernonlinearconversionoftheimagegrayscales,itispossibletohighlightspecialportionoftheimagewithcertaingrayscales.Forexample,theS-Tconversionmayeffectivelyimprovethecontrastoftheweldingdefectsandthereforethedefectmarginbecomeclear.Itsconversionequationiswrittenasperfollowing:

Where,b(x),a(x)standsforthemaximumandminimumgrayscalevalueoftherelevantdotcolumnrespectively.

BrimsSharpening

Tosharpenanimageistohighlightitsbrimswithbettercontrastforeasieridentificationoftheinternaldefectsoftheweldsorobjects.Imagesharpeningissubstantiallyatechniqueofhighpassfiltrationrealizedbyaspecialmodeloperator.Thisoperationhasobviouscharacteristicsofsimpleandefficientandthereforeisapplicableforbrimsharpening,boundarydetecting,characterizingandsmoothfilteringforthedefects.

ImageReversion

〔4〕Fortheconvenienceofthetechnicians,itisnecessarytoreversetheresultantimageanddisplayitinthesimilarwayasthenegativeimageofX-raybytheequationbelow:

Where,f(x,y),g(x,y)standsfortheoriginalandthereverseddotgrayscalevaluesrespectively.

3.2.5Pseudo-colorTreatment

Forthenumbersofthecolorsislargerthantheimagegrayscales,thegrayscalesimagecanbemirroredtotherelevantcolorswhichimprovestheidentificationorthesensitivityofhumaneyes.Suchprocessingiscalledpseudo-colortreatment.

Miscellaneous

Theotherimageprocessingapproachesincludehistogramchange,digitalfilteretc.

3.3ImageDisplayandRecording

Theprocessedimagesaredisplayedonthescreenofthemonitorforfurtherevaluationandassessment.

WeldDefectsIdentificationbyComputer

Thewelddefectcharacteristicssuchasboundarylength,longitudinaldiameter,transversaldiameter,andareasarecalculatedbytracingthedefectboundarywithcomputer.Furthermore,basedontheaboveparameters,thenameofthewelddefectwillbeidentifiedandthedimensioncalculatedbythecomputerautomaticallyasperthespecifiedempiricalformula.Forthecomputeridentificationmodelofthedefectsisverycomplicated,thedefectassessmentisstillprocessedmanuallyatpresent,butthedefectquantitativeanalysisandclassificationworkcanbecarriedoutwithcomputer.

ImageCompressioncoding

Becausethedigitalimagestakehugeamountofspace,it’snecessarytohavetheimagedatacompressedforefficienttransmissionandstorage.TheDXRTIsystemiscodedwithnon-anamorphicmeansandapplicableforthenormalcompressioncodingsuchasRLE8andLZW.

ImageStorageandFileFormat

Forthefuturecheckoutandsearch,theimagedataoftheDXRTIsystemshouldbestoredintheformatofimagefilewithlargecapacitystoragemediasuchasCD-Randkeptintheplacefreeofmagneticfield,humidity,dustandmechanicalpress.TIFFismoresuitablefortheimagefilesinDXRTIsystem.

ImageStorageDevice

Welddetectingimagesnormallystoredbytheharddiskofthecomputeratfirst,andthenaretransmittedandwrittenontoCD-RdiscbyaCD-Writer.Onemaystore4200framesofimageswithitscapacityof650MBandashelflifeofmorethan30years.Therefore,it’sthepredominantstoragemediaforthedetectedimageinformation.

4.ProgrammingforDXRTIdetectingsystem

Betweentheendof1998andthebeginningof1999,thetechniciansintheplantdevelopedanewpackageoftheDXRTIsoftwarenamed‘XCX2000’basedontheprincipleofcomputernumericalimagingandthepreviousimagingtechnique.Thesoftwarecomposedoftwoparts:theimageacquisitionprogramandimageassessmentprogramshownbytheblockdiagramsbelow.

4.1ImageAcquisitionProgram(s