Facial Expressions Classification Using Learning Vector Quantization Networks

Abstract

This paper, an automatic facial expression recognition system is developed based on Gabor wavelet methodology and learning vector quantization networks (LVQs). Facial attributes from the frontal images are extracted in the form of feature vectors by evaluation the responses from a set of 18 complex Gabor filters at 34 fiducial points on face. The resultant high dimensional feature vectors are condensed by performing PCA and are classified into classes of expression using LVQs. Since the groups of feature vectors are non-linear separable, LVQ networks are chosen for classification instead of multilayer perceptron neural networks. In this paper, the optimum number of sub-groups, learning rate and other features of LVQ are determined to maximize the recognition accuracy of the system, using a set of designed experiments.

Department(s)

Mining Engineering

Second Department

Electrical and Computer Engineering

Document Type

Article - Conference proceedings

Document Version

Citation

File Type

text

Language(s)

English

Rights

© 2004 American Society of Mechanical Engineers (ASME), All rights reserved.

Publication Date

01 Jan 2004

This document is currently not available here.

Share

 
COinS