A dynamic facial expression database for quantitative analysis of facial paralysis

Facial paralysis, the majority of which is idiopathic, causes loss of facial movement of one side. It is important to construct a facial expression databases to quantitatively evaluate this condition for early treatment of facial paralysis. Generally, the existing human facial expression databases are built using healthy volunteers for face recognition or expression recognition. In this paper, we present a newly developed dynamic facial expression database, which includes not only the facial expressions of healthy volunteers but also the facial expressions of the patients with facial paralysis. A quantitative analysis method was also proposed to investigate the application of the database. The new database can be easily extended to a 3D facial expression database for facial paralysis analysis, and is valuable for facial paralysis research and facial paralysis treatment.