Contour and Texture Analysis for Image Segmentation2

Contour and Texture Analysis for Image Segmentation2

ID:39778946

大小:2.39 MB

页数:21页

时间:2019-07-11

Contour and Texture Analysis for Image Segmentation2_第1页
Contour and Texture Analysis for Image Segmentation2_第2页
Contour and Texture Analysis for Image Segmentation2_第3页
Contour and Texture Analysis for Image Segmentation2_第4页
Contour and Texture Analysis for Image Segmentation2_第5页
资源描述:

《Contour and Texture Analysis for Image Segmentation2》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、InternationalJournalofComputerVision43(1),7–27,2001c2001KluwerAcademicPublishers.ManufacturedinTheNetherlands.ContourandTextureAnalysisforImageSegmentationJITENDRAMALIK,SERGEBELONGIE,THOMASLEUNG∗ANDJIANBOSHI†ComputerScienceDivision,UniversityofCaliforniaatBerkel

2、ey,Berkeley,CA94720-1776,USAReceivedDecember28,1999;RevisedFebruary23,2001;AcceptedFebruary23,2001Abstract.Thispaperprovidesanalgorithmforpartitioninggrayscaleimagesintodisjointregionsofcoherentbrightnessandtexture.Naturalimagescontainbothtexturedanduntexturedreg

3、ions,sothecuesofcontourandtexturedifferencesareexploitedsimultaneously.Contoursaretreatedintheinterveningcontourframework,whiletextureisanalyzedusingtextons.Eachofthesecueshasadomainofapplicability,sotofacilitatecuecombinationweintroduceagatingoperatorbasedonthet

4、exturednessoftheneighborhoodatapixel.Havingobtainedalocalmeasureofhowlikelytwonearbypixelsaretobelongtothesameregion,weusethespectralgraphtheoreticframeworkofnormalizedcutstofindpartitionsoftheimageintoregionsofcoherenttextureandbrightness.Experimentalresultsonawi

5、derangeofimagesareshown.Keywords:segmentation,texture,grouping,cueintegration,texton,normalizedcut1.Introductionagepropertiessuchasbrightness,colorandtexture.Contour-basedapproachesusuallystartwithafirstTohumans,animageisnotjustarandomcollectionstageofedgedetectio

6、n,followedbyalinkingprocessofpixels;itisameaningfularrangementofregionsthatseekstoexploitcurvilinearcontinuity.andobjects.Figure1showsavarietyofimages.De-Thesetwoapproachesneednotbethatdifferentfromspitethelargevariationsoftheseimages,humanshaveeachother.Boundari

7、esofregionscanbedefinedtobenoprobleminterpretingthem.Wecanagreeaboutthecontours.Ifoneenforcesclosureinacontour-baseddifferentregionsintheimagesandrecognizethediffer-framework(ElderandZucker,1996;Jacobs,1996)entobjects.Humanvisualgroupingwasstudiedexten-thenonecang

8、etregionsfromacontour-basedap-sivelybytheGestaltpsychologistsintheearlypartofproach.Thedifferenceismoreoneofemphasisandthe20thcentury(Wertheimer,1938).Theyiden

当前文档最多预览五页,下载文档查看全文

此文档下载收益归作者所有

当前文档最多预览五页,下载文档查看全文
温馨提示:
1. 部分包含数学公式或PPT动画的文件,查看预览时可能会显示错乱或异常,文件下载后无此问题,请放心下载。
2. 本文档由用户上传,版权归属用户,天天文库负责整理代发布。如果您对本文档版权有争议请及时联系客服。
3. 下载前请仔细阅读文档内容,确认文档内容符合您的需求后进行下载,若出现内容与标题不符可向本站投诉处理。
4. 下载文档时可能由于网络波动等原因无法下载或下载错误,付费完成后未能成功下载的用户请联系客服处理。