姿态变化与困难光照条件下的人脸识别分析.pdfVIP

ABSTRACT ABSTRACT Face isa issueintheareaof recognitionsignificant paRem the few underfrontalviewshasmade vision，etc．Inpast decades，facerecognition ratesforface under satisfactoryprogress．However，therecognition images pose variationsanddifficult conditionsstillneedtobe workfocuses lighting improved．This onface under variationsand the recognitionpose difficult conditions，and lighting contributionsareasfoIllows： 1．A Discriminative dictionarymethod，Bilinear learning Dictionary is forface under andilluminationvariations．The proposed recognition pose proposed BDDLhasthe the is on following regularization characteristics：1)the乞-normimposed makesthelearned moresuitableforface codingcoefficients，which dictionary thereconstructionerrorforeachclass the recognition．2)Byminimizing usingdictionary atomsassociatedwiththat learna make class，we structured whichisableto dictionary the reconstructionerrorforeachclassmorediscriminativefor the ofthe coefficientsof codedoverthelearned discriminability exploit coding samples discriminat