Facial expression recognition based upon human cognitive regions

The discovery of associations between local facial features and image features is one of the current challenges in regard to facial expression recognition. To mitigate the negative effects caused by individual differences, this paper proposes an approach based upon human cognition which can recognize facial expression from six human cognitive regions. The proposed approach exploits descriptions of Local Binary Pattern (LBP) to extract the holistic facial area and establishes a pseudo 3D face model for segmenting the human face into six cognitive regions. Finally, we synthesize change trends and volume distribution of LBP of the six regions to recognize the expressions. Unlike many previous studies, this approach recognizes the expressions from regional features instead of holistic facial features. Further, the regional features are more robust than interesting point-based features because of idiosyncrasy of individuals. To validate our proposed approach, we conducted experiments on the Extended CohnKanade (CK+) facial expression database. The results demonstrate that the proposed approach can effectively segment face into specific areas and recognize facial expressions.