Abstract. The analysis of facial expression temporal dynamics is of great importance for many real-world applications. Furthermore, due to the variability among individuals and different contexts, the dynamic relationships among facial features are stochastic. Systematically capturing such temporal dependencies among facial features and incorporating them into the facial expression recognition process is especially important for interpretation and understanding of facial behaviors. The base system in this paper uses Hidden Markov Models (HMMs) and a new set of derived features from geometrical distances obtained from detected and automatically tracked facial points. We propose here to transform numerical representation which is in the form of multi time series to a symbolic representation in order to reduce dimensionality, extract the most pertinent information and give a meaningful representation to human. Experiments show that new and interesting results have been obtained from the proposed approach.Keywords: Facial expression, HMM, Occurrence order, time series.
IntroductionAnalyzing the dynamic of facial features and (or) the changes in the appearance of facial features (eye, eyebrows and mouth) is a very important step in facial expression understanding and interpretation. Many researchers attempt to study the dynamic behavior. Timing, duration, speed and occurrence order in which the temporal segment of the different face/body actions occur are crucial parameters related to dynamic behavior [1]. Timing, duration and speed have been analyzed in several studies [2,3,4,5]. Little attention has been given to occurrence order [3,5]. In this paper we propose to explicitly analyze one aspect of facial dynamics by detecting the occurrence order in which facial features movement occurs. For example, expressions typically associated with happiness contain AU6 and AU12, the fact is that is not known in which order each AU occur, this is what we want to discover.As it is known, a facial expression recognition system consists generally of three steps: pre-processing, feature extraction and classification. In this paper we are not interested by the first step which relates to face localization, and registration to remove the variability due to changes in head pose and illumination.