Facial expression recognition of person without language ability based on the optical flow histogram

A method to extract the facial actions and recognize the expressions of a person without language is presented. In this proposed method, a difference image of two successive frames of an image sequence, and a corresponding I component image in YIQ image representation were combined to extract the facial action and detect the facial region. The optical flows of the corresponding image sequence, in which the facial actions appeared, were computed. Based on the optical flow projection histogram on X and Y coordinate axes in the mouth region and the eye region, the motion features of the facial expressions were extracted to compose the associate memory model of the facial expressions. By utilizing this model, the facial expressions were recognized.