基于機(jī)器視覺的紡織品外觀疵點(diǎn)檢測與質(zhì)量評價的研究.pdf

1、湖北工業(yè)大學(xué)碩士學(xué)位論文基于機(jī)器視覺的紡織品外觀疵點(diǎn)檢測與質(zhì)量評價的研究姓名：王璜申請學(xué)位級別：碩士專業(yè)：機(jī)械設(shè)計及理論指導(dǎo)教師：趙大興20090501湖北工業(yè)大學(xué)碩士學(xué)位論文●———●I!——————I■—●———————●■————■■■■●————■—●———AbstractAsproductsquailtyautomaticinspectiontechniquebasedonmachinevisionisbecomingmore

2、valuableitprovidesanimportantmethodfortherealizationofhi曲一speed，highprecisionautomaticdetectionproductsqualityFocusingontextiledefectautomaticdetectiontechniquesbasedonmachinevisionthethesisstudiesthehardwaresystemarchit

3、ectureandmajorimpactonthesystemimplementationdesignstextileimageprocessingalgorithmsanddevelopsatextiledefectdetectiondevicebasedonthesestudiesFirstofa11fordifferentdetectionrequirementsoftextileswidthandhighspeedhigh—pr

4、ecision，thecharacteristicsoftextilesintheimageacquisition，processinganddefectclassificationareconsideredcomprehensivelyEspeciallythedistributedsystemiSconducivetotheachievementofthecompletedetectionofwidetextilesurfacean

5、dCaneffectivelyreducesystemloadcalculationsAnadaptivefilterbasedonGaussiankemeliSdesignedanddevelopedafteranalyzingthetextileimagewithnoiseandtextureandthelinearfilteringmethodofdigitalimageprocessingComparingwithtraditi

6、onallinearsmoothingfilterthisfilterCaneffectivelyreducethenoiseofthetextileimageandthetargetareaoftheimagefuzzywhichiSbeneficialforthefollowingprocessingGettingdefectsregionintextileimagewithcyclicaltexture，thethesisdesi

7、gnsanimageedgeenhancementmethodbasedonadaptivefirstderivativesoperatorandmorphologicalgradient，andathresholdsegmentationoperatorwithhighprecisionselfadaptivecapacitybasedonimagestatisticsvaluesTheninorderteducethelOSSoft

8、hedetectpixelvalueafterimagepre—processing，thethesisdesignsafastdefectclusteralgorithmusingmultiscalescanningByanglicizingtextileimagesanddefectsfeatures，thethesisdesignsamethodoftextiledefectscharactersinspatialdomainan

9、dfrequencydomainextractionandsetsalibraryincludingthecharactersaStheclassificationstandard，theninputstheclassificationvaluesintotextiledefectsclassifierbasedonmendedBPneuralnetworkandgetsthedefecttypesTextilesurfacequali

10、tyevaluationsystemisdesignedbasedonChinatextileindustrystandardFZO10l8一l992andinternationalstandardISO／DIS84981990tOevaluatetextilequalityFinallyonthebasisofalgorithmsandmethodsabove—mentioned，thetextiledefectsdetectiond

11、eviceisdesignedanddevelopedTheexperimentshowsthatthedetectspeedofthedeviceiS270m／min(realtimedetection)to80~200m／rain(offlinedetection)Thehi曲estdetectionprecisioniS05mmThedeviceCandetectthedifferenttextilewidthofwhichiSf