Facial expression recognition from a partial face image by using displacement vector

Due to the existing problem of recognizing the partial face expression from a viewpoint between -/+45deg, this paper proposes a novel method for recognizing the partial face expression from a 2.5D partial face image by using displacement vector. The 2.5D partial face image is captured from a viewpoint between -/+45deg. The 3D virtual expression face is reconstructed from a 2.5D partial face image. By using face plane algorithm, the cross points are computed. And then the expression face is represented in term of displacement vectors. A displacement vector is the movement vector of a cross point pointing from a cross points of neutral face to a cross point of expression face. Finally, the displacement vectors are used for facial expression recognition by means of support vector machines. The experiments were done for four facial expressions (neutral, anger, surprise and smiling) from 22 persons, based on the assumption that the person is known. The results show the feasibility of the proposed method.