FACIAL EXPRESSION RECOGNITION USING STATISTICAL SUBSPACE

Face recognition is one of the main areas of research in computer vision. Although many studies address to, there are many challenges in this subject such as accuracy, performance, real-time applications, etc. We propose a novel model based on bilateral 2-dimensional fractional principle component analysis and examine 2-dimensional characteristic of image to retain information structure. After that, we apply statistical features to facial expression recognition problem in order to evaluate the efficiency of feature descriptor with facial images. Our proposed method is named the statistical subspace. For experiments, Cohn-Kanade dataset is used to compare the proposed model with previous methods. The empirical results show that our model is stable and efficient.