Recognizing intensity degrees from expressions in 3D facial meshes

Three-dimensional (3D) facial recognition techniques using 3D range data have been widely applied in last years. Facial expression recognition approaches using 2D static images or video-sequences have also been extensively used. However, there are few works on facial expression recognition using 3D data. Although the computational cost of processing 3D information is higher than for 2D, the problems caused by illumination, scale and pose variations are highly reduced using 3D data. In this work, the problem of recognizing intensity degrees for different types of facial expressions in facial meshes is studied. A 3D facial expression database was created to carry out the experiments. Our preliminary results show that several intensity degrees of the `anger' and `joy' expressions were successfully recognized.