资源描述：
《特征空间稳健性分析：彩色图像分割 毕业论文外文翻译》由会员上传分享，免费在线阅读，更多相关内容在学术论文-天天文库。
1、（1）外文文献原文RobustAnalysisofFeatureSpaces:ColorImageSegmentationAbstractAgeneraltechniquefortherecoveryofsignificantimagefeaturesispresented.Thetechniqueisbasedonthemeanshiftalgorithm,asimplenonparametricprocedureforestimatingdensitygradients.Drawbacksof
2、thecurrentmethods(includingrobustclustering)areavoided.Featurespaceofanynaturecanbeprocessed,andasanexample,colorimagesegmentationisdiscussed.Thesegmentationiscompletelyautonomous,onlyitsclassischosenbytheuser.Thus,thesameprogramcanproduceahighquality
3、edgeimage,orprovide,byextractingallthesignificantcolors,apreprocessorforcontent-basedquerysystems.A512512colorimageisanalyzedinlessthan10secondsonastandardworkstation.Graylevelimagesarehandledascolorimageshavingonlythelightnesscoordinate.Keywords:robu
4、stpatternanalysis,low-levelvision,content-basedindexing1IntroductionFeaturespaceanalysisisawidelyusedtoolforsolvinglow-levelimageunderstandingtasks.Givenanimage,featurevectorsareextractedfromlocalneighborhoodsandmappedintothespacespannedbytheircompone
5、nts.Significantfeaturesintheimagethencorrespondtohighdensityregionsinthisspace.Featurespaceanalysisistheprocedureofrecoveringthecentersofthehighdensityregions,i.e.,therepresentationsofthesignificantimagefeatures.Histogrambasedtechniques,Houghtransform
6、areexamplesoftheapproach.Whenthenumberofdistinctfeaturevectorsislarge,thesizeofthefeaturespaceisreducedbygroupingnearbyvectorsintoasinglecell.Adiscretizedfeaturespaceiscalledanaccumulator.Wheneverthesizeoftheaccumulatorcellisnotadequateforthedata,seri
7、ousartifactscanappear.TheproblemwasextensivelystudiedinthecontextoftheHoughtransform,e.g..Thus,forsatisfactoryresultsafeaturespaceshouldhavecontinuouscoordinatesystem.Thecontentofacontinuousfeaturespacecanbemodeledasasamplefromamultivariate,multimodal
8、probabilitydistribution.Notethatforrealimagesthenumberofmodescanbeverylarge,oftheorderoftens.Thehighestdensityregionscorrespondtoclusterscenteredonthemodesoftheunderlyingprobabilitydistribution.Traditionalclusteringtechniques,canbeusedforfeatu
