Research on Facial Expression Recognition Technology Based on Convolutional-Neural-Network Structure
Human facial expressions change so subtly that recognition accuracy of most traditional approaches largely depend on feature extraction. In this article, the authors employ a deep convolutional neural network (CNN) to devise a facial expression recognition system to discover deeper feature representation of facial expression. The proposed system is composed of the input module, the pre-processing module, the recognition module and the output module. The authors introduce jaffe and ck+ to simulate and evaluate the performance under the influence of different factors (e.g. network structure, learning rate and pre-processing). The authors also examine the anti-noise property of the system with zero-mean gaussian white noise. In addition, they simulate the recognition accuracy on different expression pairs and discuss the confusion issue on similar expression recognition. Finally, they introduce the k-nearest neighbor (KNN) algorithm compared with CNN to make the results more convincing.