正面小角度偏轉(zhuǎn)下的ASM人臉特征定位方法研究.pdf

1、廈門大學碩士學位論文正面小角度偏轉(zhuǎn)下的ASM人臉特征定位方法研究姓名：李美龍申請學位級別：碩士專業(yè)：軟件工程指導教師：段鴻20080401AbstractAbstractFacialfeaturepointlocatingintheFaceRecognition，three—dimensionalvirtualFaceSynthesis，F(xiàn)acialExpressionAnalysiSandsynthesiS，andhasaveryim

2、portantrole，iSthecurrentcomputervision，computergraphiCS，researchinthefieldofpatternrecognition，andotherhotspotsAtpresent，anumberofalgorithmsandmodelSproposed，butduetothecomplexityofthefaceandfacialimageimagingenvironment

3、alimpacts，stilldonothaveageneralmodeloralgorithmiSabletoresolvetheproblem0nInthiSpaper，thecurrentcommonfacialfeaturepointlocatingalgorithinandthemodelaresummarized，andtheirpredecessorsonthebasiSofaninitiativebasedonthesh

4、apemodel(ASM)enhancementBecausetraditionalmethodsASMdifficultpointineachfeaturearetheresultofmoreaccurate，andthetraditionalmethodofmatchingASMbytheinitialresultsoftheimpactofverylargeASMagainstthetraditionalmethodofdefec

5、tsandshortcomings，especiallyfor1ightandrobustenoughartitudechange，inordertomaketheASMmorerobustposturechangestoachievepositivedeflectionofthesmall一anglelocatingfacialfeatures。thepaperraisedanumberofimprovedmethods：(1)ase

6、lectiveandsensitivemarkeroflightpoints(eyebrows。eyes，nose)using2Dprofile(2)theuseofstackedtwoASMseriesmodelsearch，theinitialimprovementsinstalledimpropersearchfailure(3)aseparateextensionoftheeyesofthemarker，SOthatthecha

7、racteristiCSoftheeyelocatingmorerobust(4)whenusedinthesearchFaceDetectionOveratorsFaceautomaticlocatingandlocationoftheeyes，andpeoplearoundthefacewhensmall一angledeflectioncharacteristiasofamorerobustpositionExperimentalr