基于AAM算法人臉特征點的姿態(tài)估計研究.pdf

1、東南大學碩士學位論文基于AAM算法人臉特征點的姿態(tài)估計研究姓名：劉陽申請學位級別：碩士專業(yè)：軟件工程指導教師：金遠平；姚莉；李季檁2011AbstractHum鋤faceposeestimationhasbeena鑼picalhum鋤face鋤alysisproblemf0ra10ngtimeRecoveringthehumanfaceposeparameters仔omtheimageorvideo丘amesbyaspectifical

2、gorithmhumanfaceposeestimationisthekeyissuesinsolvingtheseproblemslikehum鋤computcrinteractionandfacerecognitionDomesticandforeignthehum鋤faceposeestimationmethodsc鋤besimplydividedint0threetypes：modelbasedmethod，印parenceba

3、sedmethod鋤dcl嬲sificationbasedmethodTIlrou曲thecomparision，analyizest11eadvantages鋤ddisadvantagesofthemethodsThispaperpresentsarobusthum卸紐ceposeestimationmethod：BasedonthemultiAAMtemplatesfacefeatIlresextraction，whichcanex

4、p觚dthesc叩eoffacefeaturesex仃action，鋤dcombingtheprojectionmodelt0meettllemeettherequirementofpose—vari鋤tfaceestimationInordert0designarobustfaceposeestimationalgorithmmodulesintheengineerproject，byaddingmeeliminaterestrict

5、ionsontllehumanfaCefeaturelocationin2d鋤d3dphysicalpropenies，optimizationalgorithmprocesscangiverobustaccurateresultsfinallyThereforethemethod訪thispaperc鋤giVeadecentresultinacertainrangeoffaCemultiposesituationthefinalave