World Scientific Publishing Company A SYSTEM FOR HUMAN FACE AND FACIAL FEATURE LOCATION

World Scientific Publishing Company A SYSTEM FOR HUMAN FACE AND FACIAL FEATURE LOCATION

ID:37659696

大小:576.59 KB

页数:17页

时间:2019-05-27

World Scientific Publishing Company A SYSTEM FOR HUMAN FACE AND FACIAL FEATURE LOCATION_第1页
World Scientific Publishing Company A SYSTEM FOR HUMAN FACE AND FACIAL FEATURE LOCATION_第2页
World Scientific Publishing Company A SYSTEM FOR HUMAN FACE AND FACIAL FEATURE LOCATION_第3页
World Scientific Publishing Company A SYSTEM FOR HUMAN FACE AND FACIAL FEATURE LOCATION_第4页
World Scientific Publishing Company A SYSTEM FOR HUMAN FACE AND FACIAL FEATURE LOCATION_第5页
资源描述:

《World Scientific Publishing Company A SYSTEM FOR HUMAN FACE AND FACIAL FEATURE LOCATION》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、InternationalJournalofImageandGraphicsVol.3,No.3(2003)461-479©WorldScientificPublishingCompanyASYSTEMFORHUMANFACEANDFACIALFEATURELOCATIONJUNMIAO,HONGLIU,WENGAO,HONGMINGZHANG,GANGDENG,XILINCHENICT-YCNCFRJDL,InstituteofComputingTechnology,ChineseAcademyofSciences,Beijing100080,ChinaEmail:{jmiao,

2、hliu,wgao,hmzhang,gdeng,xlchen}@ict.ac.cnThispaperpresentsanimplementationofasystemdesignedforlocationofhumanfacesandfacialfeaturessuchaspupils,eyes,noseandmouth.Thekernelofthesystemisanintegrationofseveralalgorithms,suchashumanfacecenter-of-gravitytemplate,illuminationcompensation,andsoon.Afa

3、lse-faceremovalalgorithmisproposedinthispaperespeciallyforthedistinguishingofcartoonfacesfromtruefaces.Thetestingexperimentsofthesystemshowquitegoodresults,theaveragedetectionaccuracyratesforfacedetectionandfacialfeaturelocationare97.8%and87.5%respectively.Keywords:facecenter-of-gravitytemplat

4、e,facedetection,facialfeaturelocation1.IntroductionHumanfacedetectionandfacialfeaturelocation,asspecificissuesofpatternrecognition,aresignificantinvariousapplicationssuchashumanidentification,faceretrieval,virtualhumanfacesynthesis,andmodelbasedMPEG-4facecoding.Alotofworkhasbeendoneonthisissue

5、.Proposedtechniquesincludemulti-resolution1,234analysis,fixedtemplatematching,deformabletemplatematching,center-of-gravity567templatematching,fuzzytheory,principlecomponentanalysis“eigenface”or89,10,29,3011“eigenpicture”,neuralnetwork,supportvectormachine,andusageofmultiple12,13informationincl

6、udingcolor,soundormotion.Quiteafewtechniquesarerelatedto1415facialfeatureextraction,suchasB-splines,adaptiveHoughtransform,cost1617,181920minimization,deformabletemplate,region-basedtemplate,geometricalmodel,3132linear/nonlinearfilter,activeappearancemodelsandsomeapproachesofcombined33,34sever

7、alclassicaltechniques.Faceandfacialfeaturelocationareindispensablestepsforfaceanalysis.Afast,accurateandefficientsystemforfaceandfacialfeaturedetectionishelpfulfortheimplementationofpracticalbiometricsystemsforhumanfaceanalysis.Foraprac

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。