APPLICATION OF THE KARHUNEN-LOEVE PROCEDURE FOR THE CHARACTERIZATION OF HUMAN FACES

被引：1401
作者：
KIRBY, M
SIROVICH, L
机构：
[1] Center for Fluid Mechanics, Turbulence, and Computation, Division of Applied Mathematics, Brown University, Providence
来源：
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE | 1990年 / 12卷 / 01期
关键词：
Data compression; Data extension; Face characterization; Karhunen-Loève expansion; Symmetric eigenfunctions;
D O I：
10.1109/34.41390
中图分类号：
TP18 [人工智能理论];
学科分类号：
081104 ; 0812 ; 0835 ; 1405 ;
摘要：
The exploitation of natural symmetries (mirror images) in a well-defined family of patterns (human faces) is discussed within the framework of the Karhunen-Loéve expansion. This results in an extension of the data and imposes even and odd symmetry on the eigenfunctions of the covariance matrix, without increasing the complexity of the calculation. The resulting approximation of faces projected from outside of the data set onto this optimal basis is improved on average. © 1990 IEEE.
引用
收藏
页码：103 / 108
页数：6
相关论文
共 24 条
[1]  
Abdi H., 1988, ARTIF INTELL, P149
[2]  
ANDERSON JA, 1989, PARALLEL MODELS ASS, P23
[3]  
Ash R. B., 1975, TOPICS STOCHASTIC PR
[4]  
BRUCE V, 1989, HDB RES FACE PROCESS, P487
[5]  
FUKUNAGA K, 1972, INTRO STATISTICAL PA
[6]  
Goldstein A. J., 1971, P IEEE, V59
[7]  
GONZALAZ RC, 1987, DIGITAL IMAG PROCESS
[8]   MACHINE IDENTIFICATION OF HUMAN FACES [J].
HARMON, LD ;
KHAN, MK ;
LASCH, R ;
RAMIG, PF .
PATTERN RECOGNITION, 1981, 13 (02) :97-110
[9]   RECOGNITION OF FACES [J].
HARMON, LD .
SCIENTIFIC AMERICAN, 1973, 229 (05) :70-82
[10]  
HOTELLING H, 1933, J ED PSYCHOL SEP
123