An expression recognition method based on local feature fusion
一种基于局部特征融合的表情识别方法

Expression recognition is a hot topic in artificial intelligence and pattern recognition. However, feature fusion is one of the most important research subjects in expression recognition. In this paper, an expression recognition method based on serial feature fusion between Gabor wavelet feature and geometric feature of mouth is proposed. And the nearest neighbor rule is used to classify to get the comparison from different facial expression databases and different recognition methods. The experiment results show that this method obtains better recognition effect than any other methods in local feature expression recognition. Moreover, it is proved that it can recognize express faster than global facial expression recognition.