基于計(jì)算機(jī)視覺的檢測方法與應(yīng)用研究.pdf

1、浙江大學(xué)博士學(xué)位論文基于計(jì)算機(jī)視覺的檢測方法與應(yīng)用研究姓名：漆隨平申請學(xué)位級別：博士專業(yè)：控制科學(xué)與工程指導(dǎo)教師：張宏建20050501浙江大學(xué)博士學(xué)位論文AbstractThefinaltargetofcomputervisionresearchiStomakethecomputercapableofadaptingitselftotheambientconditionbyobservingandunderstandingtheext

2、ernalenvironmentwithitsvisualsenseAtpresent，auniversalcomputervisionsystem，whichCanidentifytheobjectivesintherandombackgroundandcanindependentlyaccomplishvariousWOrkscouldnotbedevelopedButitiSpossibletodeviseaspecialcomp

3、utervisionsystemthatcarlautomaticallycompleteaspecificworkInthisPaDerseveralmeasurementapproachesbasedonthecomputervisionwerestudiedandputforwardinordertomeasurethesteelrodsonlineinthesteelandironindustryAndthenamethodfo

4、rmeasuringthesteelrodsonlinewhichintegratedinterdisciplinaryknowledgeincomputervision，probabilityandstatistics，pattemclassificationandinformationfusionhasbeendevelopedAtlastbasedonthedevelopedmethodacomputervisionsystemt

5、omeasure也esteelrodsonlineautomaticallyhasbeendevisedThemainresearchachievementisshowedasfollows：1Conceptofvirtualmultisensorwasadvancedandamethodofpatternclassificationviat11einformationfusionUSingweightedcoefficientDemp

6、sterShafer(D—S)evidencetheorywasputforwardandappliedtothecomputervisionmeasurementDiversefeaturepicturescouldbeobtainedbymeansofprocessingandtransferringthemadtimeimagewi山differentmethodsByscanningfeaturepicturesonthesam

7、epositionwithtemplatesatthesametime，degreesofbasicbeliefeachofwhichimpliesthatthereexistsameasuredobjectiveinthedetectedregioninthescannedpicturecouldbeobtained耵1enfusingthesedegreesofbasicbeliefusingweightedgoemcientD—S

8、evidencetheoryajudgmentwhetherthemeasuredobjectiveexistsatthesamepositionintheoriginalreal—timeimagecouldbegivenAfterscanningallpositioninpicturesthetotalnumbersofthemeasuredobjectiveintherealtimeimagecouldbeobtained2Inf

9、ormationfusionalgorithmusingtheselfadaptingweighedcoefficientD—Scombinationrule，anedgesdetectionmethodandaregionsegmentationmethodthattheinitialvalueoftheparametersandthresholdValHeareselfadaptivelyadjustedrespectivelyal

10、laccordingtothestatisticalcharacteristicshavebeenproposedafterstudyingthecharactersoftheonlinedisturbancesBasedontheallofabove、theselfadaptabilityofthemultisensorinformationfusionandtheantiinterfefenceabilityoftheprocess

11、toacquiretheinformationhavebeenallimproved，andthedegreesofthebasicbeliefhavebeendeterminedbythehistoricaldataandthestatisticalcharacteristicsfromtherealtimeimageThetechniqueofthemultisensormanagementiSappliedtothemeasure