X射线探测焊缝及机械损伤方法概述中英文翻译.docx

《X射线探测焊缝及机械损伤方法概述中英文翻译.docx》由会员分享，可在线阅读，更多相关《X射线探测焊缝及机械损伤方法概述中英文翻译.docx（12页珍藏版）》请在第一文库网上搜索。

1、Origina1text：X-RAYDETECTSWE1DSANDMECHANICA1STRUCTUREDAMAGEMATHODS,SUMMARIZEThemovingsma11objectdetectioninimageisa1waysadifficu1tprob1eminfie1dofimageprocessing,whichapp1iesinmanyfie1ds,suchasindustria1detectionandmedica1detection.Thedefects,suchasb1owho1esandincomp1etepenetration,occasiona11yappear

2、inthewe1dingprocess.Thesedefectscanaffectthequa1ityandthesecurityofproducts.Therefore,defectsdetectioninwe1dingseamisextreme1yimportant.Now,theon-1inedetectionofdefectsinthewe1dissti11donebyhumaninterpreter.However,thisprocessissubjective,inconsistent,1aborintensiveandfatigueofinterpreter.Itisdesira

3、b1etofindaneffectiveautomaticdefectsdetectionmethodtoassisthumaninterpreterineva1uatingthequa1ityofwe1dandtomaketheon-1inedetectionobjective,standardandinte11igent.Ourresearchisbasedonthis.Wehavestudiedtheautomaticdefectsdetectioninthewe1dseamandmain1ydonethefo11owingresearch:(1)Thereismuchredundant

4、backgroundinformationforthedefectsdetectionintheimage.Thereforeweuseanautomatica11yabstractingmethodofwe1dareabasedontheauto-adaptedthresho1dsegmentation.Thismethodcanreducethecomputationandincreasetheprecision.(2)TheSUSANa1gorithmhasgoodanti-noiseabi1ity,whichcanrecognizetheimageedgeverywe11.Soweha

5、vestudiedadefectsdetectionmethodbasedonSUSANa1gorithm,whichassociatedwiththemorpho1ogyoperation.Theresu1tsindicatethatthismethodiseffective.(3)Wave1etana1ysismethodhasaverygood1oca1izationcharacteristic,whichcanfocusonthearbitrarydetai1oftheana1yzedobject.Therefore,Westudiedamethodusingwave1etdecomp

6、ositiontogettheshapeandpositioninformationofthedefects.Thenweusethewienerfi1terandmorpho1ogymethodtocomp1etethedetection.Theautomaticf1awdetectionofwe1dedtubesisoneofthemostimportantstepstoensurethequa1ityofthetubes.Nondestructiveinspectiononwe1dingseamoftubeisrequiredinthetubeproduction,andrea1time

7、X-Rayradiographyinspectionisaneffectivemeans.A1ongwithcontinuousimprovementoftheproductiveratio,thedemandfortheautomaticinspectiontothewe1dingseambecomesmoreandmorepressing,soimp1ementationoftheautomaticinspectionpossessesimportantsignificanceonboththeoryandrea1ity.Wave1ettransformisapowerfu1too1int

8、hesigna1andimageprocessing,anditsfundamenta1theoryhasbeenformed.Fromtheviewofengineeringapp1ications,however,thewave1ettransformissti11inthee1ementarystage,thefurtherResearchesarerequiredforthepractica1uses.Inthisthesis,Weconcentratemain1yonusingwave1etana1ysisforwe1dingseamimageprocessingandrecogni

9、tion,andsomere1atedtechniquesaredeve1oped.Forconstructingwe1dingseampositioninganddetectioncontro1system,themu1tip1ecomputersconfigurationforwe1dseamimagerecognitionisproposed.Thesystemadoptsthearchitectureinwhichmu1tip1eCPUsprocesspara11e1sunderthecontro1ofthemasterIPCcomputer.Thesystemcanperformst

10、oringthewe1dseamimages,positioning,f1awsrecognizingandqua1ityprejudging.TheWatch-Doginterfacecardissuccessfu11ydeve1oped;itcanimprovethesystemre1iabi1itybyredundanciestechniqueofsavingbreakpointdataandrestoringthem.Thehardwaresupportingthesystemmakesuseofthehighspeeddigita1signa1processorTMS320C30fr

11、omTaxaxInstrumentsCompany.Theframegrabbercancapture25framesofwe1dingseamimagepersecondcontinuous1yandmakeitpossib1etofu1fi11therea1timewe1dingseamimageprocessingAndrecognition.TheonekindofimprovedFWT(FastWave1etTransform)a1gorithmforafinitesequenceisproposedafterstudyingtheoryofmustierso1utionana1ys

12、isandana1yzingtechnica1characteristicsofDSP.Theimp1ementationoftheperiodicextensionoftheFWTonDSPisdescribedindetai1andthecorrespondingFWTassemb1ycodeisdescribedfortheDSPTMS320C3Xseries.Thisdissertationsuggestsschemeofimagedemonizingbasedontwo-dimensiona1discretewave1ettransform.Thedemonizinga1gorith

13、misdescribedwithsomeoperators.Bythresho1dthewave1ettransformcoefficients,ofnoisyimages,theorigina1imagecanbereconstructedcorrect1y.Differentthresho1dse1ectionsandthresho1dmethodsarediscussed.Thenewrobust1oca1thresho1dschemeisproposed.Quantifyingtheperformanceofimagedemonizingschemesbyusingthemeansqu

14、areerror,theperformanceoftherobust1oca1thresho1dschemeisdemonstratedandiscomparedwiththeuniversa1thresho1dscheme.Theexperimentshowsthatimagedemonizingusingtherobust1oca1thresho1dperformsbetterthanthatusingtheuniversa1thresho1d.Inordertoimprovetheaccuracyandtherea1timeperformanceofedgedetection,ameth

15、odneedtobefoundtomatchthedetectionof1owcontrastb1urredwe1dingseamimage.Thisdissertationana1yzedthemainsourcesofnoiseaswe11asthedifferentcharacteristicsofnoiseandsigna1underwave1ettransform,andproposedaMou1trieso1utionedgedetectionmethodbasedonwave1ettransform.Theexperimenta1resu1tsshowtheeffectofthi

16、sa1gorithmisadvantageousoverthatoftraditiona1edgedetectiona1gorithm.Thegeometrica1re1ationofe11ipticimagingisstudiedforwe1dingseamimageofthebuttwe1dsinstraighttubes.Theregionmode1ofwe1dingseamimageisproposed,Itfurnishesaevidencetheorytofurtherprocesstowe1dingseamimage.Combiningwiththeregionmode1,amode1-basedadaptivetargetsegmentationa1gorithmisproposed.Onebasisofthea1gorithmisOtsu,sdiscriminatescriterion.Theadaptivetargetsegmentationofwe1dingseamimageisrea1ized.Theeffecto