教学文库网 - 权威文档分享云平台
您的当前位置:首页 > 文库大全 > 专业资料 >

基于皮肤模板和改进HMM的自动人脸识别系统

来源:网络收集 时间:2026-08-04
导读: 基于皮肤模板和改进HMM的自动人脸识别系统 第25卷第1期2008年1月 深圳大学学报理工版 JOURNALOFSHENZHENUNIVERSITYSCIENCEANDENGINEERING VnL25No.1 Jan.2008 文章编号:1000

基于皮肤模板和改进HMM的自动人脸识别系统

第25卷第1期2008年1月

深圳大学学报理工版

JOURNALOFSHENZHENUNIVERSITYSCIENCEANDENGINEERING

VnL25No.1

Jan.2008

文章编号:1000-2618(2008)01.0071.05

【电子光学与信息工程】

Automaticfacerecognitionbased

andimproved

SHENLin—iinand

on

skinmasking

HMM

Zhong

MING

CollegeofSoftwareShenzhenUniversityShenzhen518060P.R.China

Abstract:AnewhiddenMarker

module(HMM)basedfacerecognitionsystemispresentedinthispaper.Face

imageswereextractedautomaticallyfromlivevideoscapturedbyrealtimeusing

all

CreativeWebCam,andfaceswererecognizedin

improvedHMMfacerecognitionalgorithm.AfastfacedetectionalgorithmusingboostedHaar-like

re-

featureswasappliedinitiallytodetectfacesregionsfromthevideostream.Thedetectedfaceregionwasfurtherfinedby

skincolormaskingmoduletoachievemoreaccuratefaceposition.ToimprovetheaccuracyofHMMbased

facerecognitionalgorithm,discretewavelettransform,insteadofdiscretecosinetransform,wasusedtoextractob—servationsequencesforHMM。Experimentswereconductedusingtwofacedatabases:theORLdatabaseandtheNottinghamcolorfaceimagedatabase.Theresults

accuracy

on

bothdatabasesshowthattheproposedmethod

can

improvethe

bymorethansixpercents.Furtherimprovementhasbeenobservedwhenskinmaskingmoduleisusedto

refinethedetectedfaceKeywords:face

region.

detection;colorimageprocessing;biometrics;hiddenMarkovmodel

recognition;face

CLCNumber:TP391:TP751

Automated

Documentcode:A

facerecognitionsystemshave

as

widerangeofim—thesystem.Itsaccuracyismask

as

improved

using

to

fastskincolorbased

portantapplicationssuchvideosurveillance,crimeprevention

postprocessingprocedure

to

locatethefaceareaaccu-

anddetection.Acompletefacialimagerecognitionsystemshouldbeable

sanle

rately.Duefaces

Call

the

accuracyofthedetectedfaces,thedetectednormalized.

to

to

detectfacesin

givenimageand

recognize

thematthewill

beeasily

time.Oncethefacepositionis

fromthe

determined,facial

inboth

area

Major

approaches

face

recognitionincludegraphmate—

be

extracted

image,normalized

acaleandoften-

hing,neuralnetworksandhiddenMarkovmodels(HMM).An

thedynamiclinkarchitecturefirst

tation,andpassed

onto

thefacerecognizer.

Rowley'sneural

of

exampleofgraphmatchingis

Atypicalfacedetectionsystemis

basedfacenetworks

networl【-ofneural

proposedbyVonderMalsburg。“.Morerecentlythefacebunch

graph‘“iswavelets

detector[1|.Thedetectorconsists

by

set

proposedapplied

toat

copewithvariationsinfaceimages.Gabor

manuallyselectedftducialseveralimagesof

trained

largetraining

set

offaceimagesand

non-

are

points(eyes,

the

faceimages.Violarithmusingface

et

al[2J

proposedrapid

objectdetectionalgo-

cascadeofface/non -

mouth,nose,etc.)of

refereed

to

as

faceandtheresults.

set

ofHaar likefeatureand

jeta,arepackedin

…… 此处隐藏:1657字,全部文档内容请下载后查看。喜欢就下载吧 ……

基于皮肤模板和改进HMM的自动人脸识别系统.doc 将本文的Word文档下载到电脑,方便复制、编辑、收藏和打印
本文链接:https://www.jiaowen.net/wenku/268230.html(转载请注明文章来源)
Copyright © 2020-2025 教文网 版权所有
声明 :本网站尊重并保护知识产权,根据《信息网络传播权保护条例》,如果我们转载的作品侵犯了您的权利,请在一个月内通知我们,我们会及时删除。
客服QQ:78024566 邮箱:78024566@qq.com
苏ICP备19068818号-2
Top
× 游客快捷下载通道(下载后可以自由复制和排版)
VIP包月下载
特价:29 元/月 原价:99元
低至 0.3 元/份 每月下载150
全站内容免费自由复制
VIP包月下载
特价:29 元/月 原价:99元
低至 0.3 元/份 每月下载150
全站内容免费自由复制
注:下载文档有可能出现无法下载或内容有问题,请联系客服协助您处理。
× 常见问题(客服时间:周一到周五 9:30-18:00)