An expressional dynamic feature extraction method based on Gabor filtering

Using existing dynamic feature extraction method to extract expression features from image sequence, the facial appearance features are also extracted together. Aiming at this problem, an expressional dynamic feature extraction method based on Gabor filtering is proposed in this paper. Expression features being extracted, the method takes advantage of the frequency and direction selectivity of Gabor filter to suppress the facial appearance feature extraction. Thereby facial appearance features are reduced in expression features extracted from the sequence. The experimental results in Cohn-Kanade face database and CMU-AMP face database show expression features extracted by the proposed method are much more efficient for expression recognition of image sequence.