基于改進梯度函數(shù)的圖像質(zhì)量評價及l(fā)aplace算子邊緣檢測新方法的研究

1、I分類號：密級：學號：2006092單位代碼：10407碩士學位論文論文題目:基于改進梯度函數(shù)的圖像質(zhì)量評價及基于改進梯度函數(shù)的圖像質(zhì)量評價及LaplaceLaplace算子邊緣檢測新方法的研究算子邊緣檢測新方法的研究研究方向圖像處理圖像處理專業(yè)名稱計算機應用技術(shù)計算機應用技術(shù)研究生姓名陳志輝陳志輝導師姓名、職稱劉小生劉小生教授教授20082008年1010月1010日江西贛州江西贛州IIIAbstractWiththeextensiv

2、eapplicationofdigitalimagevideotheissueofdigitalimagequalityassessmentbecomesmemeimptant.Asabasicproblemofimageprocessingimagequalityassessmentisnotonlyimptantintheybutalsowantedinwidebackgroundsofapplication.Digitalimag

3、equalityassessmentusuallycanbedividedintosubjectiveassessmentobjectiveassessment.Subjectiveassessmentmethodistedioustimeconsumingcannotbeimplementedinrealtime.Objectiveassessmentmethodsareclassifiedintothreecategies:Full

4、Referencequalityassessment(FRQA)ReducedReferencequalityassessment(RRQA)NoReferencequalityassessment(NRQA).Thefmertwotypesofresearcharefullydevelopedthelatterisarelativelynewfield.AsthedevelopmentofthehumanvisualsystemThe

5、rehavebeenthreetypesofimagequalityassessmentmethod:imagequalityassessmentbasedonHumanVisualSystem(HVS)imagequalityassessmentbasedontheconcernedfactsimagequalityassessmentbasedonstructuraldisttion.Theimagequalityassessmen

6、tbasedontheconcernedfactsthefactscanbetteraffecttheimagequalitydisttionthereasonstheresultsoftheassessmentcanbetterreflectthehumanvisualexperience.Thisdissertationintroducestheimptanceofimageinfmationthedefinitionofimage

7、qualityassessmentsummarizescomparessometypicalimageassessmentmethodsatpresentanalyzestheacteristicsofHumanVisualSystemshowstheimptancenecessityofconsideringHVSinimagequalityassessment.Consideringconcernedfactsthedisserta

8、tionproposesanewmethodofimagequalityassessment.AccdingtotheacteristicofHVSextractsthreeindicatsthatinfluenceimagequalityi.e.brightnessclarityrelevance.Thenthedissertationintegratesthethreeindicatsgetstheimagequality.This

9、methodfullyintegratestheacteristicsofHVSthebrightnessindicatisconsistentwiththeexpressioninspatialdomaintheclarity(blur)referstotheproptionofhighfrequencycomponentsinthefrequencydomainusingImprovedGradientFunctiontoexpre

10、ssion.TheImprovedGradientFunctionisthroughtheeightdirectionsoftheeachofpixelofimagetoachievethevalueofthegradient.relevanceindicatisusingthesimilarityoftwoimagestoexpression.Evaluatethedifferentofdisttionofimageobtainthe

11、weightofthreefactsthroughtheexperiment.UsingtheweightofthreefactsevaluatetheimageofknowtypesofdisttiontheresultscanbetterreflecttheHumanVisualSystem.Themethodsnotonlyaffecttheimagequalityoftherelevantfactstheacteristicso