资源描述：
《基于多特征的彩色图像融合分割方法研究》由会员上传分享，免费在线阅读，更多相关内容在行业资料-天天文库。
1、合肥工业大学硕士学位论文基于多特征的彩色图像融合分割方法研究姓名：张利利申请学位级别：硕士专业：信号与信息处理指导教师：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
