资源描述:
《基于机器视觉的嵌入式仪表读数识别系统分析-(5695)》由会员上传分享,免费在线阅读,更多相关内容在行业资料-天天文库。
1、浙江理工大学硕士学位论文基于机器视觉的嵌入式仪表读数识别系统研究提升了系统的安全性。最后,本文对系统进行集成和测试,实验结果表明,该系统能够很好地满足实际需求。关键词:嵌入式识别终端;数字识别;指针识别;机器视觉;并行化任务;加密算法II万方数据浙江理工大学硕士学位论文基于机器视觉的嵌入式仪表读数识别系统研究ResearchofEmbeddedMeterReadingRecognitionSystemBasedonMachineVisionABSTRACTAtpresent,restrictedbyworkingconditions,economics
2、andconvenience,inmechanicalmetrologyfield,thereisaheavyuseofmechanicalandanaloginstrument,whosereadinganddatarecordingworkcurrentlycanonlybeartificialcompleted.Inordertoimproveworkefficiency,reducethemanualinterventionandreducetheerrorprobability,instrumentvisualdetectionsystemw
3、ithPCcenteredisoftenused.Althoughtheyreducetheburdenofartificialreadingtoacertainextent,theystillhavethedisadvantagesoflargevolume,highcost,beingnoteasilyandrapidlydeployedandidlecomputingpower.Thispaperdesignsanddevelopsasetofembeddedmeterreadingrecognitionsystembasedonmachinev
4、isiontechnology,whichcanbeautomaticallyreadandconvertedexistingpointertypeanddigitalwatermeterdial,toreachautomaticidentificationanddatastorage.Thehardwaresystemmainlyreferstotheembeddedhardwareterminalidentificationsystem.Thesubsystemisbasedonembeddedprocessor,andisequippedwith
5、industrialcamerawithhighresolution.Thehardwaresubsystemismainlyresponsiblefortheacquisitionofimagedata,therealizationofreadingrecognitionalgorithm,andprovidingthenecessaryman-machineinterfaceetc.Aimingatthesupportvectormachine(SVM)’sdisabilitytohavefastrecognitionspeedandthehigh
6、recognitionrateatthesametimeinpracticalapplication,thispaperputsforwardanewidentificationalgorithmbasedonimagepixelgradientchanges.Thispaperwouldtakethenaturalgasmeterasthesampletable,dopositioning,pretreatment,extractionofdigitalfeaturepointclassificationaswellasaseriesofoperat
7、ionstoitsdialdigitalregion.Meanwhile,thispaperconstructsaclassificationmodelbasedonSVMdigitalfeatures,usescrossvalidationideatooptimizeparameters,andatlastprovidesasetofschemeguaranteeingboththerecognitionrateandidentificationspeed.Besides,thispaperalsoputsforwardanimprovedpoint
8、ertyperecognitionalgorithmaimedataIII万方数据浙江理工大学