Inspection system for rail surfaces using differential images

《Inspection system for rail surfaces using differential images》由会员上传分享，免费在线阅读，更多相关内容在学术论文-天天文库。

1、Page1of92017-METC-0713InspectionSystemforRailSurfacesUsingDifferentialImagesF.J.delaCalle,DanielF.Garcia∗,RubenUsamentiaga´∗∗DepartmentofComputerScienceandEngineering,UniversityofOviedo,CampusdeViesques33204Gijon,Asturias,Spain,Email:fjcalleherrero@gmail.com´Abstract—Inthis

2、paperasurfaceinspectionsystemforrailsisThispaperpresentsanewsurfaceinspectionsystemforproposed.Thereisalackofcommercialsystemsandpublicationsrailsbasedontheprocessingofdifferentialimagesobtainedaboutsurfaceinspectionofrails.Therefore,thequalitycontrolfromthesurfaceofrailsby

3、aspectralimagedifferentiationdevelopedbyrailmanufacturersdependsontheirowndevelopedprocedure.systemsorontheequipmentsprovidedbyveryfewsellers.CommercialsystemsmustbeconfiguredmanuallybyexpertsTherestofthispaperisorganizedasfollows:inSectionIIofsellercompanies.Theconfiguration

4、processrequireslargesomeofthemostrelevantrelatedworksaboutdefectdetectionsetsofalltypesofmanufacturedrailsalongseveralperiodsareintroduced.InSectionIIIaglobalvisionoftheproposedoftimeandtheactiveparticipationofthequalityengineerssystemisgiven.InSectionsIVandVmethodsofacquis

5、itionofthemanufacturer.Thisisalong,cumbersomeandveryandprocessingofimagesaredescribed.InSectionVIaexpensiveprocess.Inthispaperweproposeanewsystemthatcanbeconfiguredusingasystematicmethodthatcanbemethodforoptimallyconfiguringtheparametersofthesystemperformedbythequalityenginee

6、rsofthemanufacturer.Theisdescribed.FinallySectionsVIIandVIIIgiveexperimentalproposedinspectionsystemusesdifferentialimagesoftherailresultsandconclusions.surfacesobtainedwithatechniquecalledtheSpectralImagesDifferentiationProcedure.TheseimagesareprocessedusingaII.RELATEDWORK

7、computervisionalgorithmthatlooksforvariationsinthepixelvaluesamongtheimages.Inordertooffermoreinformation,aSurfaceinspectionbasedoncomputervisiontechniquesisneuralnetworkapproachisusedforclassifyingdetecteddefectssuitablefordiversekindofmaterialsandindustries.Manyintosixtyp

8、es.Theproposedsystemisanopensolutionforthepapersstudythiskindofimageprocessingalgo