Automatic face parts extraction and facial expression recognition

Real-time facial expression analysis is an important yet challenging task in human computer interaction. This paper proposes a real-time person independent facial expression recognition system using a geometrical feature-based approach. The face geometry is extracted using the modified active shape model. Each part of the face geometry is effectively represented by the Census Transformation (CT) based feature histogram. The facial expression is classified by the SVM classifier with exponential χ² weighted merging kernel. The proposed method was evaluated on the JAFFE database and in real-world environment. The experimental results show that the approach yields a high recognition rate and is applicable in real-time facial expression analysis.