Fusion of Multiple Facial Features for Age Estimation

被引:1
作者
Lu, Li [1 ]
Shi, Pengfei [1 ]
机构
[1] Shanghai Jiao Tong Univ, Inst Image Proc & Pattern Recognit, Shanghai 200240, Peoples R China
基金
中国国家自然科学基金;
关键词
age estimation; Gabor wavelets; (2D)(2)PCA; SVM;
D O I
10.1587/transinf.E92.D.1815
中图分类号
TP [自动化技术、计算机技术];
学科分类号
0812 ;
摘要
A novel age estimation method is presented which improves performance by fusing complementary information acquired from global and local features of the face. Two-directional two-dimensional principal component analysis ((2D)(2)PCA) is used for dimensionality reduction and construction of individual feature spaces. Each feature space contributes a confidence value which is calculated by Support vector machines (SVMs). The confidence values of all the facial features are then fused for final age estimation. Experimental results demonstrate that fusing multiple facial features can achieve significant accuracy gains over any single feature, Finally, we propose a fusion method that further improves accuracy.
引用
收藏
页码:1815 / 1818
页数:4
相关论文
共 10 条
[1]  
[Anonymous], FG NET AGING DATABAS
[2]  
CHANG CC, LIBSVB LIB SUPPORT V
[3]  
Gao Wen., 2004, The cas-peal large-scale chinese face database and evaluation protocols
[4]   Automatic age estimation based on facial aging patterns [J].
Geng, Xin ;
Zhou, Zhi-Hua ;
Smith-Miles, Kate .
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, 2007, 29 (12) :2234-2240
[5]   Image-based human age estimation by manifold learning and locally adjusted robust regression [J].
Guo, Guodong ;
Fu, Yun ;
Dyer, Charles R. ;
Huang, Thomas S. .
IEEE TRANSACTIONS ON IMAGE PROCESSING, 2008, 17 (07) :1178-1188
[6]   Age classification from facial images [J].
Kwon, YH ;
Lobo, ND .
COMPUTER VISION AND IMAGE UNDERSTANDING, 1999, 74 (01) :1-21
[7]  
LANITIS A, 2002, 14 INT C DIG SIGN PR
[8]  
Vapnik Vladimir, 1999, The nature of statistical learning theory
[9]   2D Gaborface representation method for face recognition with ensemble and multichannel model [J].
Wang, Lin ;
Li, Yongping ;
Wang, Chengbo ;
Zhang, Hongzhou .
IMAGE AND VISION COMPUTING, 2008, 26 (06) :820-828
[10]   An approach for directly extracting features from matrix data and its application in face recognition [J].
Xu, Yong ;
Zhang, David ;
Yang, Han ;
Yang, Jing-Yu .
NEUROCOMPUTING, 2008, 71 (10-12) :1857-1865