Facial Expression Recognition using DWT-PCA with SVM Classifier

Over the last two decades, Facial Expression Recognition is an important field in Computer vision & Artificial Intelligent. Facial Expression plays an important role in Human communication, so it is mostly used in Human Computer Interaction (HCI) interface. Recognizing Human Facial Expression is challenging task because of the variation in illumination, pose and occlusion. In this study,we introduce a hybrid face recognition technique, consisting of two main parts namely feature extraction and classification. In the first part, as feature extracting techniques, we benefit from Eigenfaces method which is based on Principal Component Analysis (PCA) and Discrete Wavelet transform (DWT). In the second part, after generating feature vectors, Support Vector Machines (SVMs) are utilized. We examined the classification accuracy according to three different SVM kernel types. For the test set, we focused on Jaffe database including 213 images of 10 different females. At the end of the overall recognition task, we have obtained the classification accuracy 96.67% with Radial Basis Function (RBF) kernel.