DEFORMABLE QUATERNION GABOR CONVOLUTIONAL NEURAL NETWORK FOR COLOR FACIAL EXPRESSION RECOGNITION

In facial expression recognition (FER) tasks, convolutional neural networks (CNNs) have been shown their great capability of learning features. In this paper, we describe a new framework for FER in color images, which incorporates deformable Gabor filters into quaternion CNNs. Deformable Gabor filters reinforce the network’s ability of extracting different orientations of facial wrinkles information. Quaternion CNNs have greater advantages over the regular CNNs in handling the coupling between the color channels. The proposed deformable quaternion Gabor convolutional neural network (DQG-CNN) not only learns FER feature representation excellently, but also processes spectral correlation between color channels naturally. Moreover, it can also effectively reduce the training complexity compared to other reference models. Experimental results on three benchmark color datasets Oulu-CASIA, MMI, and SFEW, demonstrate that proposed DGQ-CNN outperforms some other state-of-the-art methods clearly.