Contribution of Facial Transient Features in Facial Expression Analysis: Classification & Quantification

The understanding of an image deals with the consideration of the most banal information conveyed by each pattern. A crucial anchor for facial expression analysis is to consider not only permanent features but transient ones too. In this work we are interested by all the outcomes of transient features in the analysis of facial expressions. Three processes based on transient features are proposed. The first process is proposed to classify a facial expression as one of the six universal expressions. The presence or absence of transient features on different facial regions is studied to associate each transient feature to a specific facial expression. The result is generally a doubt between two or more than two expressions. A post processing step is added to reduce doubt if exists between expressions. The second process is proposed to classify expressions into negative, positive or unknown classes. In this process, transient features are associated to negative or positive expressions. The angle formed by the nasolabial furrows (if it exists) came to complete the classification of the considered expression in the two classes. Finally a process to quantify an expression with nasolabial furrow comes to conclude the outcomes of transient features in facial expression analysis, where the angle formed by the nasolabial furrow is calculated to deduce if the expression intensity is “High” or “medium”. Obtained results are given to prove the reliability of each process.