Face Recognition via Two Dimensional Locality Preserving Projection in Frequency Domain
In this paper we investigate the face recognition problem via using the two dimensional locality preserving projection in frequency domain. For this purpose, we first introduce the two-dimensional locality preserving projections (2DLPP) and the two dimensional discrete cosine transform (2DDCT). Then the 2DLPP in frequency domain is proposed for face recognition. In fact, the 2DDCT is used as a preprocessing step and it converts the image signal from time domain into frequency domain aiming to reduce the effects of illumination and pose on face recognition. Then 2DLPP is applied on the upper left corner blocks of the global 2DDCT transform matrices of the original images, which represent the central energy of original images. For demonstration, the Olivetti Research Laboratory (ORL), YALE and FERET face datasets are used to compare the proposed approach with the conventional 2DLPP and 2DDCT approaches with the nearest neighborhood (NN) metric being used for classifiers. The experimental results show that the proposed 2DLPP in frequency domain is superior over the 2DLPP in time domain and 2DDCT in frequency domain.
Two Dimensional Locality Preserving Projections Two Dimensional Discrete Cosine Transform Face Recognition
Chong Lu Xiaodong Liu Wanquan Liu
School of Electronic and Information Engineering DLUT, Dalian, 116024, China YiLi Normal College, Yi School of Electronic and Information Engineering DLUT, Dalian, 116024, China Curtin University of Technology, Perth WA, 6102, Australia
国际会议
无锡
英文
271-281
2010-09-17（万方平台首次上网日期，不代表论文的发表时间）