Face Recognition by Subspace Analysis of2D Log-Gabor Wavelets Features
In this paper,we discuss a face recognition schemeby subspace analysis of 2D Log-Gabor waveletsfeatures.In which,an input face image is firstlydecomposed with a set of two dimensional Log-Gaborwavelets(2D-LGWs)localized with respect to spatiallocation,orientation and frequency.Based on complexresponses of filters,local energy model(LEM)is usedto represent Log-Gabor features(LGFs)which aresubstantially effective for the task of recognition.Then,subspace modeling is performed to transform the highdimensional LGFs into more compact one to simplifythe task of classification.Common nearest-neighbor(NN)based matching algorithm is adopted to classify aprobe to one of classes.The superiority of theproposed scheme for face recognition is comparativelydemonstrated with the traditional appearance-basedmethods.Moreover,performances of several leadingsubspace techniques,PCA,ICA and LDA,arecomparatively evaluated based on LGFsrepresentation.
Ling Fan Hong Duan Fei Long
Tan Kah Kee College,Xiamen University,Zhangzhou 363105,China Software School,Xiamen University,Xiamen 361005,China
国际会议
厦门
英文
1167-1172
2008-11-17（万方平台首次上网日期，不代表论文的发表时间）