学位论文
Abstract
Withthecontinuousdevelopmentofimageprocessingtechnology,imageprocessingtechnologyhavebeenwidelyusedinthenationaleconomyandallaspectsofsociallife.Atthesametime,therequirementsoftheimageprocessingisincreasing.TraditionaldigitalimageprocessingdevicesaremainlyASIC(ApplicationSpecificIntegratedCircui0andDSP
bythe(DigitalSignalprocessors).Sincethebeginningofthe20thcentury,accompanied
developmentofsemiconductortechnology,withitsapplicationofflexible,highlyintegrated,
Gatepowerful,shortdesigncycle,andlowdevelopmentcosts,FPGA(FilmProgrammable
Array)areincreasinglybeingusedinthefieldofimageprocessing.Moreandmorepractice
that,FPGA'sparallelprocessingcapabilityandpipelineoperationshasprovedCan
dramaticallyimprovethespeedofimageprocessing,theFPGA-basedimageprocessingsystemhasbroadprospectsfordevelopment.
Inthispaper,weresearchedaimagepre-processingsubsystemin
aanembeddedvideomonitoringsystem.First
andofall,werealizedhardwareplatformforcommonre-configurableimageprocessingalgorithmwhichcouldcompleteimageacquisition,receiving,processing,storage,outputotherfunctions.FPGAitselfentirelyisaresultofre—configurable,SOthe
structureofthehardwareplatformcall
usedthebeeasilyupgradedandrepeatlyconfigurated.Second,toachieveavarietyof
tointheplatform,thisarticleVerilogHDLhardwarelanguageimagepre-processingalgorithmintheFPGAchip.Intherealizationoftheprocess,inorder
givefullplaytotheFPGA’Sstrongpowerinparallelprocessing,wedo
improvementtomakeitpossibletouseacertainamountofparallelprocessingmethods.Theexperimentalresults
ashowthattheimagepre—processingsystemisabletocompletevarietyofhigh—levelimage
algorithmsinmilliseconds,whichcompletelymeetthe50/Soutputrequirementsofthevideosurveillancesystem.
Finally,accordingtoavarietyoftypesofnoisearisingfromthepracticalapplicationofvideosurveillancesystem,wedesignedaimageprocessingcontrolmodulebasedonfeedbacktheory.ThemoduleCananalysethepeaksignaltonoise
aratio(PSNR)oftheimageshandled,thencontrolFPGAtousingmoretargetedapproachtoprocessthenextimage.
Keywords:imageprocessing,FPGA,parallelprocessing,noise,PSNRII