資源描述：
《基于多特征的彩色圖像融合分割方法研究》由會(huì)員上傳分享，免費(fèi)在線閱讀，更多相關(guān)內(nèi)容在行業(yè)資料-天天文庫。
1、合肥工業(yè)大學(xué)碩士學(xué)位論文基于多特征的彩色圖像融合分割方法研究姓名：張利利申請學(xué)位級別：碩士專業(yè)：信號與信息處理指導(dǎo)教師：胡良梅2011-04Colorimagefusionsegmentationmethodsbasedonmulti-featureABSTRACTImagesegmentationisanimportanttechnologyinimageprocessing,andhasbeenwidelyappliedintraffic,medicine,agriculture,indust
2、ry,andsoon.Itisalsoaclassicalproblemincomputervisionwhichhasnotbeenwellsolved.Therearemanymethodsofimagesegmentation,however,mostofthemonlycanbeusedinspecificimages,donothaveuniversalapplicabilityandvalidity.Inrecentyears,someresearchersintroduceinfor
3、mationfusionstrategytoimagesegmentationinordertoimprovethesegmentationeffect.TheresearchofcolorimagefusionSegmentationmethodhasbroadapplicationprospects.Thisthesisincludesthefollowingcontents:(1)Weintroducedthedefinitionandthegeneralprocessofimagesegm
4、entation,summarizedthemainmethodofimagesegmentation.(2)Weintroducedthedefinitionandbasicprincipleandhierarchicalstructureofinformationfusion,summarizedthemethodofimagesegmentationbasedonfusion,describedthesegmentationmethodbasedonfeaturefusionandthese
5、gmentationmethodbasedonmulti-scalefusionindetail,collatedtheclassicalevaluationcriteriaofimagesegmentation.(3)Inordertosolvetheproblemofpresentingcomplexsceneinformationinaspecificsinglecolorspace,weuseamethodbasedonhierarchicalclusteringtofusemultipl
6、esegmentationresultsofmultiplecolorspace.WecarriedoutsegmentationexperimentsonBerkeleysegmentationdatabaseandcomparedwithavarietyofclassicalsegmentationmethods.Theexperimentalresultsindicatedthatthissegmentationmethodcangethighersegmentationaccuracy,a
7、ndhadadvantagesforovercomingoversegmentationproblem.(4)AccordingtothesegmentationevaluationcriteriaPRI,wederiveafusionmodelforcombiningmultiplesegmentationresults,throughminimizingtheGibbsenergyfunctionofthismodeltoobtainetheoptimalfusionresult.Wecarr
8、iedoutavarietyofexperimentsonBerkeleysegmentationdatabaseandcomparedwithotherclassicalsegmentationmethods.Theexperimentsindicatedthatthesegmentationresultsofthismethodweremoreconsistentwiththegroundtruth.Keywords：Imagesegmentation;Fusionsegmen
