Facial asymmetry for expression recognition
GE Wan-cheng · Information technology newsletter · 2009
An expression recognition algorithm based on facial asymmetry is described.The algorithm represents the facial asymmetry with AsymFaces(D-face,Sface).The facial asymmetry is selected automatically according to the augmented variance ratio(AVR) and improved variance ratio.The classification rate is measured with the Euclidean distance and KNN(K-Nearest Neighbor).The experiment results show that the facial asymmetry contains discriminative information and its application in expression recognition can improve the performance of expression recognition algorithms.