基于粗糙集與混合特征的人臉表情識別研究.pdf

1、江蘇大學(xué)碩士學(xué)位論文基于粗糙集與混合特征的人臉表情識別研究姓名：段麗申請學(xué)位級別：碩士專業(yè)：計算機(jī)應(yīng)用技術(shù)指導(dǎo)教師：張建明20091215江蘇大學(xué)碩士學(xué)位論文ABSTRACTTheFaceExpressionRecognition(FER)referstousethecomputertechnologytoextractfeaturesfromthefaceexpressioniimageaccordingtothehumanthink

2、ingwaytoclassifythefeatures，inordertOanalyzeandunderstandthehumanfeelingsfromthehumanfaceinformationTheFERisachallengingcrosscuttingissuewhichisrelatedtobiologicalfeaturesrecognition，pattemrecognition，imageprocessing，mac

3、hinevision，movementtracking，physiologypsychologyandSOonAtthesametime，itisanimportantpartofaffectivecomputingandintelligenthuman—computerinteractionwhichhasbroadapplicationsandpotentialmarketvaluesInthisthesis，wefirstlydi

4、scussthebackgroundandthenanalyzethemainexitingexpressionfeatureextractionandrecognitionalgorithmsOnthisbasis，weproposethemethodofFacialexpressionRecognitionBasedonRoughSetandFusedFeature。nemaincontentsaredescribedasbello

5、ws：(1)TheroughsetisintroducedtoFER，andthealgorithmofimprovedroughsetattributereductionisproposed，itmakesuptheshortcomingsoftheoriginalattributereductionalgorithmwhichonlyembarksfromthekernelattributesItconsiderstheimport

6、ancedegreeoftheattributetakesintoaccountthecontributionsdegreeofvariousattributesSoitcanbeusedtoselectexpressionfeatureseffectively(2)ToovercometheproblemofexistingWPCAwhichonlystressedsomeortwegioninformationandweighted

7、regionsshapeisnotflexibleenough，theupdatedweightprincipalcomponentsanalysisalgorithm(tn胛CA)isproposedItscattersemphasisontheregionswhichhaveimportantcontributionstotheexpressionchangessuchastheeye，eyebrowandmouth，80thefa

8、ceexpressionfeaturesaremoreprominentThen，twocoefficientsofthehorizontalandverticaldirectionarejoinedtoachievetwowayadjustableoftheweightregion(3)Inordertoobtainmoleeffectivefeature，anewmethodbasedonthepartialfeaturesandt

9、heoverallfeaturesisproposedtoextractthehybridfeaturesAAMmethodisusedtopartialposition，select32keypointsfromtheeyebrows，eyes，nose，cheekandmouthareas，calculatethedifferentdistancesbetweendifferentpoints，Usetheroughsettosel

10、ectthefeaturesrtoachievethepartialgeometryfeatures；UWPCAisusedtotheoverallfeaturesfromtheexpressionimageswiththeroughsetThentheKCcAisusedtofusethesetwofeaturesastheobservationvectorsofdiscreteHMM，whichcanachieveagoodresu

11、lt(4)AprototypesystemoffacialexpressionrecognitionbasedonroughsetandfusedfeatureisdesignedandimplementedThemodulesinthissystemmainlyinclude：Imagepretreatment，expressionfeatureextraction，featureselection，emotionfeaturefus