Abstract
This paper investigates emotion recognition from brain waves measured by using a (Cyberlink)^<(TM)> which acquires and processes biological potentials from his/her forehead. Based on an emotional dimension theory in psychology, 2D-emotional plane that includes "pleasure-unpleasure" axis and "arousing-sleepy" axis is introduced As a first step of emotion recognition, relationship between the arousal level and EEG is discussed and a classification method of the arousal level is presented. The averaged EEG power of α and β wave bands constructs a pattern vector and an artificial neural network is utilized as a classifier. Using the collected EEG data from Japanese male subjects, the classifier is trained by using SOM and LVQ and tested. The obtained rate of the arousal-relax classification is 90% and the classification rate of arousal level is 60%.
