A Bayesian approach to recognise facial expressions using vector flows

: Facial expressions play an important role in human nonverbal communication. They can be generated by activation and dilatation of facial muscles. In this paper we describe a system to recognize facial expressions automatically. Special areas in the face have been selected to extract features from the vector flow of visual muscle activity. To classify facial expressions Bayesian Networks have been used. The classifier has been trained and tested on video recordings from the Cohn Kanade database. It contains recordings from the six basic emotions as defined by Ekman. The model and results of testing are reported in the paper. Key words: Bayesian networks, Automatic Recognition of Facial Expressions, Vector Flow, Cohn Kanade database.

• Type: Journal
• Year: 2009
• Where: COMPSYSTECH
• Authors: Xiaofan Sun, Léon J. M. Rothkrantz, Dragos Datcu, Pascal Wiggers
• Keywords: Cohn Kanade Database, COMPSYSTECH 2009, Facial Expressions, Hardware, Vector Flow