22652.基于分數(shù)階偏微分方程的圖像去噪算法研究

1、獨創(chuàng)性聲明本人聲明所呈交的論文是我個人在導師指導下進行的研究工作及取得的研究成果。盡我所知，除了文中特別加以標注和致謝的地方外，論文中不包含其他人已經(jīng)發(fā)表或撰寫過的研究成果，也不包含為獲得寧夏大學或其它教育機構(gòu)的學位或證書而使用過的材料。與我一同工作的同志對本研究所做的任何貢獻均已在論文中作了明確的說明并表示了謝意。研究生簽名：佬移時間：7口q年歲月79日關(guān)于論文使用授權(quán)的說明本人完全了解寧夏大學有關(guān)保留、使用學位論文的規(guī)定，即：學校有

2、權(quán)保留送交論文的復印件和磁盤，允許論文被查閱和借閱，可以采用影印、縮印或掃描等復制手段保存、匯編學位論文。同意寧夏大學可以用不同方式在不同媒體上發(fā)表、傳播學位論文的全部或部分內(nèi)容。(保密的學位論文在解密后應(yīng)遵守此協(xié)議)研究生簽名：惺衫時間：矽仰年籮月耀日導師簽名：舞且時間：知心年rR坊J5lAbstractWiththedevelopmentofthehumansocietyandthesciencetechnologythedigit

3、alimageprocessinghasbeenwidelyappliedtovariousfields，fromtheverybeginningoftheaviationindustryandtheinformationofcommunicationstomodemindustrybiologymedicineofagriculture，intelligentrobotsindustryandmanyotheraspectsNoise

4、interferenceintheprocessoftransmissionandacquisitionofdigitalimagesreducesimagequalityandinfluencessubsequentprocessing，suchasimagecompressionandimagesegmentationDenoisingalgorithmsarecommonlyusedindigitalimageprocession

5、，includingmeanfiltering，lowpassfilteringandWienerfiltering，etcHoweverwhenthealgorithmsmentionedaboveareusedtosuppressnoise，fuzzyprocessingistakentotheedgesofimagesThepreservationofthedetailedinformationandtheremovalofthe

6、noiseinformationiSacontradictionthatcannotreconcileinthedenoisingprocessesThetraditionaldenoisingalgorithmswouldunderminetheimageedge，textureandotherdetailswhenremovingnoiseThedenoisingalgorithmbasedonpartialdifferential

7、equations，whichCanmakeselectivesmoothingfortheimage，anditisabletesolvethiscontradictionperfectlyThepartialdifferentialequationsimagedenoisingmodelwasresearchedbythecombinationoftheoreticalanalysisandsimulationexperiments

8、inthispaperThemainresearchworkofthispaperisdividedintothefollowingcontents：Severalclassicalimagedenoisingmethodsbasedonpartialdifferentialequationsareintroduced，Analyzingthesemodelsandsummarizingtheiradvantagesanddisadva

9、ntagesOnthebasisoftotalvariationdenoisingmodel，andtwonewimprovedmodelsareproposedInthispaperdoingthesimulationexperimenfortheabovedenoisingmodelsbyMATLABt001And，analyzingtheexperimentalresultsbysubjectiveandobjectiveeval

10、uationmethodsoftheimagequaiity(1)BycombiningfractionaldifferentialoperatorwhichcanenhanceimagetextureinformationwithvariationalpartialdifferentialequationandthenapplyingtoimagedenoisingAdenoisingmodelbasedonfractionalpar

11、tialdifferentialoperatorisputforwardThemodelCannotonlybettersuppressnoiseoftheimage，butalsobetterpreservedetailedtextureinformationHowever，theorderofthefractionaldifferentialoperatormustbedeterminedbyalotofexperimentsThe

12、refore，calculatinglocalvarianceischosentoexpressthecomplexityofthelocalimagetextureSOastorealizeadaptivedeterminationofthe0rderoffractionaldifferentiation(2)FordisadvantagesofthefidelitytermcallblurtheimageedgeProposingt

13、healgorithmwhichcombinethegradientfidelitytermwithfractionalorderanisotropicdiffusionIntroducingtheconceptoflocalgradientfidelitySOthatweightedthegradientfidelityterm，usealgorithmofgradientfidelitytermfordenoisinginthepa