基于面陣CCD的冷軋帶鋼表面缺陷在線檢測關(guān)鍵技術(shù)研究.pdf

1、重慶大學(xué)碩士學(xué)位論文基于面陣CCD的冷軋帶鋼表面缺陷在線檢測關(guān)鍵技術(shù)研究姓名：楊根齊申請學(xué)位級別：碩士專業(yè)：機械電子工程指導(dǎo)教師：湯寶平20090523重慶大學(xué)碩士學(xué)位論文英文摘要ABSTRACTColdrolledstripsarethemainproductsofsteelindustryandalsothechiefrawmaterialofmanyotherindustriesAsmoreandbeRerrequestsfrom

2、customers，moreandmoreintenselymarketcompetition，inspectingthesurfaceofthecoldrolledstriphasallgreatimportanteffectonimprovingthestrip’SqualityandstrengtheningmarketcompetitionforacompanyThereforeresearchinganddevelopingo

3、n—linecoldrolledstrips’surfaceinspectionsystemhasbeenanimportantsubjectathomeandabroadInviewoftheproblemsincurredintheon—lineinspectionsystemofcoldrolledstripbasedonareaCCD，suchasinsufficientrealtimedataprocessing，lowdef

4、ectrecognition，someintensivestudyhavebeenconductedandsomemethodsareproposedtosettletheseproblemsMainworksofthethesisareasfollows：①Inviewofthememoryoccupiedbymassdataduringcameraacquiringandthefluctuatingscreencausedbythe

5、mismatchbetweentheacquiringspeedandtheprocessingspeed，theimagereal—timeprocessingmethodjointthemulti—threadandcirclequeueispurposedTheimageacquiringthreadandimageprocessingthreadaredetachedSOastomakethemworkingsynchronou

6、slyThesparetimeofCPUisutilizedfullytheproblemofimagedataaccumulationiseffectivelysolved，andtherealtimeperformanceofthesystemisimprovedeffectively②Inviewofthepoorantinoiseperformanceoftraditionalcameraselfcalibration，anal

7、yzingtheprincipleofcameraselfcalibrationbasedoncircle—point，makessomeimprovementsbasedonthetraditionalAnewcameraselfcalibrationapproachWasproposedTheexperimentalresultsdemonstratethismethoddoesverywellonthetheoreticsimpl

8、icitytherobustnessandveracityoftheresult③Themassiveimagesarefastdiscriminatedintodefect—uncertainimagesandno—defectimages，usingthefastinspectingmethodbasedonmultivariatediscriminantfunctionOnlythedefectuncertainimageswil

9、lbeprocessed，mostofimageswerefilteredandtheaccountofdatamustbeprocessedweredecreasedthroughthismethodTherealtimeperformancewasfurtherassuredandthedefectdetectionwasimproved④AfterComparingandresearchingtheperformanceofsev

10、eralclassicalalgorithmsoffilteringandsegmentingoperatorsandtheapplicabilitytostripsurfaceimages，itfindsthatthemedianfiltercanfilterthepulsenoiseeffectivelybutalsokeepstheimagefigureandthespeedisalsofastAlthou曲thedefectbo