基于自組織神經網絡和模糊邏輯的圖像處理.pdf

1、天津理工大學碩士學位論文基于自組織神經網絡和模糊邏輯的圖像處理姓名：孫震申請學位級別：碩士專業(yè)：信號與信息處理指導教師：王兆霞201206AbstractDigitalimageinpaintinghasbecomeapromisingresearchfieldinmultipixelima2eprocessingareasince2001Ithasawiderangeofapplicationssuchastherestoration

2、ofdamagedimagesorphotosandtheremovalandreplacementofanoptionalareaItsprimarygoal1StorepairthedamagedareaandmakeitlooksbetteraccordingtohumanvisualcharacteristicsandthecharacteristicsofthesurroundingpixelsTheconventionali

3、mageinpaintingmethodCanberoughlydividedintotwokinds0neisappliedtoinpaintthedamagedregionsinasmallscale，theotheroneisemployedtor印airthedamagedregionsinalargerscaleTheformerismainlybasedonpartialdifferentialequations(PED)a

4、ndthelatterisusuallythetexturebasedcompletionmethodWiththedevelopmentofcomputertechnology，avarietyofintelligentmethodsteplacetraditionalmethodsofimageprocessing，soItcomputingiSatypicalrepresentativeSoftcomputingincludesa

5、rtificialneuralnetworks(ANN)，fuzzylogic(FL)andgeneticalgorithm(GA)Thelowcostsolutionscanbeobtainedbyusingthesesoftcomputingtechniquestodealwiththeuncertain，impreciseandincompletetruevaluefauItt01eranceThesesoftcomputingt

6、echniquesaremutuallycomplementaryInthispaperwewillcombineneuralnetworkandfuzzylogictoinvestigateimageinpaintingThispaperstudiesthetraditionalimageinpaintingmethods，analyzesthetheoreticalprinciples，andgivesthesimulationre

7、sultsofthemostrepresentativemethodsAftergainingexperiencesfromthetraditionalmethods，thispaperproposesnewinpaintingmethodsandgivessimulationresultsThemaincontributionsofthispaperareasthefollowing：1)Soft。computingtechnique

8、sareemployedtostudydigitalimageinpaintingAnimageinpaintingmethodbasedontheselforganizingmap(SOM)andK—meansalgorithmisproposed2)AnimageinpaintingmethodbasedonfuzzylogicisproposedThemeritsandeffectivenessofaboveproposedmet