Ambiguous Data in Automated Facial Expression Recognition

PUBLISHED:
ASCI'99 Proceedings of the fifth annual conference of the Advanced School for Computing and Imaging, Heijen, The Netherlands, June 15-17, 1999, pp. 125-132.

KEYWORDS:
Hybrid facial feature extraction, knowledge-based reasoning on ambiguous facial data, facial action tracking, facial expression emotional classification.

ABSTRACT:
This paper discusses Integrated System for Facial Expression Recognition (ISFER), which performs recognition and emotional classification of a facial expression from a still facial image. The system consists of two major parts. The first one is the ISFER Workbench, which forms a framework for a hybrid facial feature tracking. The second one is the Human Emotion Recognition Clips Utilised Expert System (HERCULES), designed to convert a low level face geometry into a high level face action description, and then this into the highest level weighted emotion labels. The currently available image-processing techniques do not facilitate a completely accurate facial features detection. On the other hand, HERCULES has been developed to reason on a set of exact facial data [19]. The development of the ISFER Workbench according to a multi-detector paradigm and employment of the data redundancy to define unambiguous face geometry forms the main topic of this paper.