Facial expression recognition algorithm based on feature fusion adaptive weighted HLAC

In this paper, a novel method of facial expression recognition based on adaptive weighted higher-order local autocorrelation coefficient (WHLAC) is presented. Firstly, The method gets whole face region and sub-regions of eyebrows, eyes, nose and mouth. Secondly, the global features of the face regions and local features of the sub-regions are extracted by HLAC, the weights of sub-regions are calculated by fisher linear discriminant (FLD), and then these two parts features are fused together by the weights. Finally, the fused features are classified by FLD. The experimental results show that the proposed method has higher recognition rate and lower computational cost than Gabor, WLBP, HLAC for facial expression recognition.