An Active Snake Model for Classification of Extreme Emotions

The paper presents an image based method for classification of extreme emotions. Facial images were analyzed using action units derived from the distance-measures, taken from the centers of eyes and the mouth. After initial edge detection, the contours of the mouth and the eyes were further traced and smoothened using an active snake model. Two states of emotions i.e. happy and sad have been tested in this paper. The ratio of the mouth opening and the average distance between the center of mouth and the eyes has been considered as the indicator of the extreme emotional states.