An exploration toward automatic facial expression analysis for intelligent HCI systems
Jun Hakura, Masaki Kurematsu, Hamido Fujita · 2007
This paper describes our ongoing exploration toward facial expression analysis for intelligent HCI systems. Two approaches are proposed from two different points of views. The first approach captures dynamic movement accompanied with facial expressions as the clue for estimating emotion. In this approach, the movements are considered as the results of activities by virtual systems that are assumed to be embedded in the face. We call this approach as System Identification Approach. The second approach relies on the positions of the facial parts, such as eyes, eye brows, and mouth, to detect emotion. This approach, named Facial Expression Map Approach, relates the positions of the parts with the six basic emotions and the relations are represented as maps. The outlines of the methods based on the approaches are described with some experimental results.