Facial expression recognition using angle-related information from facial meshes

In this paper we introduce a new method for facial expression recognition. In order to be able to recognize the six main facial expressions [1] we use a grid approach and therefore we establish our new feature space based on the angles that each grid´s edge form. This way we undertake several affine transformations such as translation, rotation and scaling which in other approaches are considered very harmful in the overall accuracy of a facial expression recognition algorithm. We will therefore demonstrate how we create this feature space, as well as how we apply a feature selection process within this space. The angular nature of the data impose some considerations which will be clarified in this paper.