Classification of Facial Images Using Gaussian Mixture Models
Lists
Only you see this
Abstract
We present a new technique for face recognition. Two distinct and mutually exclusive classes of difference between two facial images are defined: within-class differences set (differences in appearance of the same individual) and between-class differences set (differences in appearance between different individuals). Then Gaussian mixture models (GMMs) are used to estimate the eigenspace densities of the two classes. And subsequently a matching similarity measure is computed based on the maximum likelihood (ML) method. The new method achieved as much as 45% error reduction compared to the standard eigenface approach on the ORL database.
BibTeX
@inproceedings{liao2001classification,
author = {Pin Liao and Wen Gao and Li Shen and Xilin Chen and Shiguang Shan and Wenbing Zeng},
title = {{Classification of Facial Images Using Gaussian Mixture Models}},
booktitle = {IEEE Pacific Rim Conference on Multimedia},
year = {2001},
doi = {10.1007/3-540-45453-5\_93},
}Loading related content...