Deep neural network for face recognition based on sparse autoencoder

Face recognition is a very important research topic in computer vision because of its many potential applications. In this paper, we investigated a face recognition method based on deep neural network. The sparse coding neural network and the softmax classifiers were used in this paper to build and train the deep hierarchical network after the face image preprocessing. The method is evaluated on the ORL, Yale, Yale-B and PERET face database, respectively. The experimental results show that the deep learning method can abstractly express the original data with efficiency and accuracy, and achieve a good performance in the conditions of illumination, expression, posture and low resolution.