Significance of facial features in performance of automatic facial expression recognition

Abstract

Automatic facial expression recognition is a fascinating and challenging problem, and impacts important applications in many areas such as human-computer interaction (HCI) and robotics. In this paper an automatic facial expression recognition method is proposed, we are applying face detection methods to an image from the dataset to get face image and its important parts like eyes, nose and mouth automatically. Local binary patterns are used as feature extractor and for classification a strong machine learning classification tool support vector machine is used. Our experiments illustrate that the LBP provide a compact and discriminative facial representation and by adopting Support Vector Machines we obtained the best recognition performance of 95.83% on Cohn-Kanade database, which is better than contemporary methods. We experimentally illustrate that eyes and mouth play a significant role in facial expression recognition.