Skip to Main Content
Present methods of facial expression recognition usually designate an expression image as one kind of six facial basic expressions. However, a facial expression usually is a complex expression that consists of several basic expressions. This paper proposes a facial complex expression recognition algorithm based on fuzzy kernel clustering and support vector machines. This algorithm designs the binary facial complex expression classification tree by using fuzzy kernel clustering algorithm, trains support vector machines at each node of the binary classification tree and describes the complexity of a facial expression according as the result of support vector machines classification. Experimental results indicate that the proposed algorithm generates higher accuracy for the JAFFE database and achieves better performance than 1-a-r SVMs. In addition, experimental results show that the result of the proposed method is more accord with practice than the result of traditional expression recognition methods.