Facial Expression Recognition Using NKFDA Method With Gabor Features

Face representation based on Gabor features is an effective method to extract facial appearance information for the advantages of the Gabor filters. However, Gabor features currently adopted by most systems are redundant and too high dimensional. In this paper, we develop a novel facial expressions recognition approach using AdaBoosted Gabor features, which are not only low dimensional but also discriminant. The method consists of three modules. First, face detection is used. Second, Gabor wavelet and AdaBoost are applied to select feature that presents a face image. Finally, Gabor features selected by AdaBoost are fed into NKFDA to classify. Experiments on Cohn-Kanade database demonstrate that these Gabor features selected by AdaBoost are enough to achieve good performance comparable to that of methods using bmp images directly