It is generally known that human faces, as well as body motions and gestures, provide a wealth of information about a person, such as age, race, sex and emotional state. This monograph primarily studies the perception of facial expression of emotion, and secondarily of motion and gestures, with the purpose of developing a fully automated visual affect recognition system for use in modes of human/computer interaction. The book begins with a survey of the literature on emotion perception, followed by a description of empirical studies conducted with human participants and the construction of a 'face image database'. On the basis of this work, a visual affect recognition system was developed, consisting of two modules: a face detection subsystem and a facial expression recognition subsystem. Details of this system are demonstrated and analyzed, and extensive performance evaluations and test results are provided. Finally, current research avenues leading to visual affect recognition via analysis of body motion and gestures are also discussed.
