An effective feature extraction method for facial expression recognition using adaptive Gabor wavelet

Feature extraction is an important and challenging phase of facial expression recognition problem. In this paper, an effective feature extraction method is proposed. Our facial feature representation method is based on an adaptive Gabor wavelet transform. In this method, we used a fuzzy controller for tuning the orientation parameter of filter. This filter can detect the most significant edges of facial images. Furthermore, the proposed adaptive filter improves the drawbacks of conventional Gabor filters. Nearest neighbor and multi-class Support Vector Machine (SVM) classifiers are applied for classification task. Experimental results on Japanese Female Facial Expression (JAFFE) database show that the proposed method can provide high recognition rate. The main advantage of proposed method over other methods is its flexibility.