Affective Computing Technology Based On Dynamic Facial Expression Recognition

Facial expression recognition is a foundation in affective computing. And facial expression is produced by activities of facial muscles, so the spatial and temporal relations between different facial muscles are very important in the facial expression recognition process. However, these implicit relations have not been widely used due to the limitation of the existing models.In order to make full use of spatial and temporal information, a special Bayesian network named as Interval Algebra Bayesian Network is introduced to capture the temporal relations among facial events. The corresponding algorithm for facial expression modeling and recognition is also developed. The proposed method does not require the peak frames and only uses the features based on tracking results, which can improve the speed of training and recognition. Experimental results on the benchmark databases CK+ and MMI show that the proposed method is feasible in facial expression recognition and considerably improves the recognition accuracy. The recognition rates in the two databases are increased to 88.1% and 62.5% respectively.Based on the above dynamic expression recognition method, we finally design a harmonious human-computer interaction prototype system. In the system, the virtual human can recognize users’ expression from camera, and then respond appropriately. The reaction of virtual human to user is communicating with user using different words according to the user’s current emotional state, such that relieve user’s sentiment and make him relax. For instance, if the user’s facial expression is identified as sad, the virtual human will tell a joke to make the user happy. Being tested again and again, the system is proved to be highly available and effective.