New research advances of facial expression recognition

As an important part of the technology for human-machine interface, facial expression recognition (FER) draws much attention recently and numerous methods are proposed. In this survey we present the facial expression recognition research and characteristics which focus on the development in the past five years. A new classification method of feature extraction is proposed in this paper, through which we can realize the development trend of feature extraction methods. Currently, the automatic and multi-scale methods of feature extraction are widely used. With the rapid development of the technology of three-dimensional model, some novel methods are used to feature extraction. At the same time, improved expression classification methods (ANN, SVM, HMM) are emphasized. Lastly, we demonstrate the state and possible challenges of the technology of FER and provide some valuable advices about the development trend of FER.