文档下载 文档下载导航
设为首页 | 加入收藏
搜索 请输入内容:  
 导航当前位置: 文档下载 > 所有分类 > IT/计算机 > paper1220_final
侵权投诉

paper1220_final

快速Mean-shift聚类

Paci cGraphics2010

P.Alliez,K.Bala,andK.Zhou(GuestEditors)

Volume29(2010),Number7

Ef cientMean-shiftClusteringUsingGaussianKD-Tree

ChunxiaXiao

MengLiu

TheSchoolofComputer,WuhanUniversity,Wuhan,430072,China

Abstract

Meanshiftisapopularapproachfordataclustering,however,thehighcomputationalcomplexityofthemeanshiftprocedurelimitsitspracticalapplicationsinhighdimensionalandlargedatasetclustering.Inthispaper,weproposeanef cientmethodthatallowsmeanshiftclusteringperformedonlargedatasetcontainingtensofmillionsofpointsatinteractiverate.Thekeyinourmethodisanewschemeforapproximatingmeanshiftprocedureusingagreatlyreducedfeaturespace.Thisreducedfeaturespaceisadaptiveclusteringoftheoriginaldataset,andisgeneratedbyapplyingadaptiveKD-treeinahigh-dimensionalaf nityspace.Theproposedmethodsigni cantlyreducesthecomputationalcostwhileobtainingalmostthesameclusteringresultsasthestandardmeanshiftprocedure.Wepresentseveralkindsofdataclusteringapplicationstoillustratetheef ciencyoftheproposedmethod,includingimageandvideosegmentation,staticgeometrymodelandtime-varyingsequencessegmentation.

CategoriesandSubjectDescriptors(accordingtoACMCCS):I.4[Computingmethodologies]:ImageProcessingandComputerVision—Applications

1.Introduction

Meanshiftisawellestablishedmethodfordatasetclus-tering,ithasbeenwidelyusedinimageandvideoseg-mentation[CM02][WTXC],objecttracking[CRM03],im-agedenoising[BC04],imageandvideostylization[DS02][WXSC04],andvideoediting[WBC 05],italsohasbeenextendedtogeometrysegmentation[YLL 05]and3Dre-construction[WQ04].Meanshiftworksbyde ningaGaus-siankerneldensityestimateforunderlyingdata,andclusterstogetherthepointsthatconvergetothesamemodeundera xed-pointiterativescheme.Althoughmean-shiftworkswellfordataclusteringandobtainpleasingclusteringre-sults,however,thehighcomputationalcomplexityistheoneofmaindif cultiestoapplymeanshifttoclusterlargedataset,especiallyforthosesituationswheretheinteractiveandevenrealtimeclusteringprocessingarepreferred.

ThecomplexityforthestandardmeanshiftprocedureisO(τdn2),wherenisthenumberofthedatapoints,theτisthenumberoftheiterationsformeanshiftcluster-ingprocedure,anddisthedimensionofthepoint.Themostexpensivecomputingisto ndtheclosestneighbor-hoodforeachpointinthedataspace,whichisamultidi-mensionalrangesearchingmethod.Evenusingoneofthe

c2010TheAuthor(s)

c2010TheEurographicsAssociationandBlackwellPublishingLtd.Journalcompilation

PublishedbyBlackwellPublishing,9600GarsingtonRoad,OxfordOX42DQ,UKand350MainStreet,Malden,MA02148,USA.

mostpopularnearestneighborsearchmethod,theANNmethod[AMN 98],givenaquerypointqandε>0,a(1+ε)approximatenearestneighborofqmustbecom-putedinO(cd,εlogn)time,wherecd,εisafactordepend-ingondimensiondandε.Therefore,whenprocessinglargedatasets,thehightimecomplexityleadstoseriousdif culty.Althoughmanyaccelerationtechniqueshavebeenproposed[EDD03,GSM03,YDGD03,WBC 05,CP06,PD07,FK09],furtherimprovementsarestilldesirableforbothperfor-manceandclusteringquality.

Inthispaper,inspiredbythefasthigh-dimensional lter-ingmethodusingGaussianKD-trees[AGDL09],wepro-poseanef cientparadigmformean-shiftprocedurecom-puting.Ourmethodisbasedonfollowingkeyobservation,sincethemeanshiftprocedureclustersthosepointsthatarefeaturesimilar,whiletherearemanyclustersofpointswhicharehighsimilarinfeature,itiswastefultoperformmeanshiftprocedureforeachpointtoconvergetothemode.Thuswe rstclustertheoriginalpointsetintoclustersbasedonfeaturesimilarityusingKD-tree[AGDL09],andobtaintheGaussianweightedsamplesoftheoriginaldataset,whichisthereducedfeaturespaceforapproximatingoriginalpointset.Theninsteadofcomputingmeanshiftdirectlyonorigi-

第1页

你可能喜欢

热门文档

相关文档

  • 3080_Final_2012_answers

    Answers Final Exam CHEM 3080 April 9, 2012, 2 pm-5 pm CB 121 Name ... no additional paper, but you are allowed to use the back of the pages...

  • final paper

    Academic paper Final Paper Literature review ? including the backgrand, previous research limitation, purpose, method, findings and results, futhur research, ...

  • final_draft

    final_draft_其它语言学习_外语学习_教育专区。The Final Draft Reflecting on ...8 ? by 11 inch white paper ? double spaced, size 12 font ? one inch...

  • 120703_UNIQLO PR Plan CN final-14时 (NXPowerLite)

    120703_UNIQLO PR Plan CN final-14时 (NXPowerLite)_销售/营销_经管营销_...paper (release) ● 媒体/政府关系 ● 准备并进行面对媒体的准备 ● 紧急时...

  • final-exam paper

    Appendix Final exame paper of systemic anatomy for foreign studnts in WZMC Part One: 5.5 Years MBBS Major, 1 Semester of 2006-2007, Systemic Anatomy...

  • ...Multiprocessor_Architecture_white_paper_Final_v1...

    Tegra_Multiprocessor_Architecture_white_paper_Final_v1.1 tegra2 nvidiategra2 nvidia隐藏>> Whitepaper NVIDIA Tegra Multi-processor Architecture V1.0 P age...

  • 邱媛媛_小微_final

    1000-1220(2010)07- An approach for Building Complex Network of Web Service...In this paper, the approach of building complex network of web service is...

  • English_Final_Examination_Paper_A

    English_Final_Examination_Paper_A_英语考试_外语学习_教育专区。山东建筑大学往年期末及答案College English Achievement Test (Band I ) 试卷类型 A Paper One ...

  • take-home paper final

    take-home paper final_英语学习_外语学习_教育专区 暂无评价0人阅读0次下载举报文档 take-home paper final_英语学习_外语学习_教育专区。Yintong Liu ...

  • 西门子面试题 A6_Answers_to_paper_testing_Final

    西门子面试题 A6_Answers_to_paper_testing_Final_IT/计算机_专业资料。A6答案Answer to paper testing of SCM Excellence Graduate Program 2011 I. 翻译 1. ...

站点地图 | 文档上传 | 侵权投诉 | 手机版
本站所有资源均来自互联网,本站只负责收集和整理,均不承担任何法律责任,如有侵权等其它行为请联系我们.
文档下载 Copyright 2013 doc.guandang.net All Rights Reserved.  -  email