PUBLISHED:
Conference Proceedings of EUROMEDIA '98, Leicester, United Kingdom,
January 5-6, 1998
KEYWORDS:
ABSTRACT:
Human Expression Recognition Clips Utilised Expert
System [Pantic, 1996] was designed to recognise facial expressions of the
observed person in an automatic way. HERCULES, recognition and interpretation
part of ISFER system [Rothkrantz et al, 1998], forms a part of an Automated
System for Non-verbal Communication [van Vark et al, 1995] and represent
the implementation of the facial expression analysis that the system performs.
The theoretical formulation of facial expression recognition, whereupon
inference engine of HERCULES is built, has been acquired from FACS [Ekman,
1978]. HERCULES contains also a second inference engine - emotion recognition
engine that offers an emotional interpretation to the recognised facial
expression. Currently HERCULES' emotion recognition is based on the recognition
of so-called six basic emotions defined by Ekman [Ekman, 1975, 1982]: happiness,
sadness, fear, surprise, disgust and anger. Making the system independent
of any of the propagated psychological theories about emotions will be
the next step in the development of HERCULES. The system now accepts manually
made measurements on a full-face photograph and returns a description of
the shown facial expression, quantitative measurement of it, its interpretation
in terms of the six basic emotions and quantitative measurement of that
emotional interpretation. Although the current version of HERCULES operates
with a manual input, a fully automated version of the system is in its
latest phase [Rothkrantz et al, 1998].