Faical Expression Recognition by Combining Texture and Geometrical Features

Most of the existing facial expression recognition methods are based on either only texture features or only geometrical features. In this paper, we propose to improve the performance of facial expression recognition by combining both types of features using fuzzy integral. The geometric features used are the displacements of positions of feature points on the face. We first embed them in a lower dimensional manifold space, then use a modified version of Support Vector Machine (SVM) as the classifier. The texture features are boosted Gabor features. Since the dimension of Gabor features is quite high, we use Adaboost to select the most important features and then use SVM to classify them for different emotions. Finally, we combine these two methods using fuzzy integral. The experiment results show that our method significantly improves the performance of facial expression recognition.