Color Face Recognition by Using Quaternion and Deep Neural Networks

In this paper we propose a new model for color face recognition based on quaternion number and deep neural networks (DNN), to enhance the classification accuracy of color face recognition. The proposed model is derived by introducing image represented in quaternion domain as an input vector in DNN model, frequently, utilized in many vision tasks of pattern recognition. The utilization of quaternion representation (QR) aims to reduce the size of input vector and consequently the computational complexity. On the other hand, color information from the image while obtaining best classification results. The simulations results are performed on color face databases to demonstrate the effectiveness of the proposed model. The obtained results show that they outperform other existing algorithms.