Layered Fuzzy Facial Expression Generation: Social, Emotional and Physiological

Abstract

Facial expression plays an important role in human’s daily life, as indicated by (Mehrabian, 1968), in face-to-face human communication, only 7% of the communicative message is due to linguistic language, 38% is due to paralanguage, while 55% of it is transferred by facial expressions. Currently, facial expression has been widely researched in psychology, sociology, cognitive science, biology, pathology, neuroscience, computer science, and so on, thus different views of facial expressions have been formed. Facial expression has been researched in the “Emotion View” for a long time. In the 19th century, kinds of facial expressions of emotions were studied by (Darwin, 1872), who argued that there is a link between emotions and expressive behaviour. Later, facial expressions of 6 basic emotions or fundamental emotions were pointed to be universally recognized in the cross culture studies by (Ekman & Friesen, 1971). However, the assumption of universality of human facial expressions of emotion was suggested to be premature by (Russell, 1994). As there are some shortcomings in the emotion view (Fridlund, 1997), the “Behavioral Ecology View” treats facial displays as social signals of intent. Facial display may depend upon the intent of the displayer, the topographic features of the niche, the behaviour of the recipient, and the context of the interaction (Fridlund, 1994). Recently, facial expressions have been considered as emotional activators and regulators (Lisetti & Schiano, 2000). It has been found that voluntary facial action can generate subjective experience of emotion and emotion specific autonomic nervous system activity. Along with the rapid development of the research fields of computer science, humancomputer interaction, affective computing, etc., the generation of facial expressions in computer has been actively researched. The Improvisational Animation system (Perlin, 1997) generated facial expressions such as angry, daydreaming, disgusted, distrustful, fiendish, haughty etc. by relating lower level layer facial movements to higher level moods and attitudes. (Yang et al., 1999) proposed a facial expression synthesis system, in which 34 facial expressions were generated by converting emotion into combination of upper, middle and lower expressions. The CharToon system (Hendrix & Ruttkay, 2000) generated kinds of facial expressions by interpolation between the 7 known expressions (neutral, sadness, happiness, anger, fear, disgust, and surprise) positioned on the emotion disc. (Bui et al., 2001) realized a fuzzy rule-based system, in which the animated agent’s representations of 7 single expressions and blending expressions of 6 basic emotions were mapped onto muscle