By Topic

A hybrid facial expression recognition method based on neutral face shape estimation

Sign In

Cookies must be enabled to login.After enabling cookies , please use refresh or reload or ctrl+f5 on the browser for the login options.

The purchase and pricing options are temporarily unavailable. Please try again later.
2 Author(s)
Sezer Ulukaya ; Elektrik-Elektronik Mühendisliği Bölümü, Boğaziçi Üniversitesi, Turkey ; Çiğdem Eroğlu Erdem

In order to recognize the facial expression of a person, the knowledge of the neutral facial expression of that person is useful but may not always be available.We present a method based on Gaussian mixture models (GMM) to estimate the unknown neutral facial expression of an expressive face. The estimated neutral face is then subtracted from the features of the expressive image and classified using support vector classifiers (SVC). Experimental results on the extended Cohn-Kanade (CK+) database give an emotion recognition rate of 88% using geometric features only and 92% if appearance based features are also included.

Published in:

2012 20th Signal Processing and Communications Applications Conference (SIU)

Date of Conference:

18-20 April 2012