Facial Shape Spaces from Surface Normals and Geodesic Distance

Abstract

The analysis of shape-variations due to changes in facial expression and gender difference is a key problem in face recognition. In this paper, we draw on ideas from the field of statistical shape analysis to construct shape-spaces that span facial expressions and gender, and use the resulting shape-model to perform face recognition under varying expression and gender. Our novel contribution is to show how to construct shape-spaces over fields of surface normals rather than Cartesian landmark points. According to this model face needle-maps (or fields of surface normals) are points in a high-dimensional manifold referred to as a shape-space. The similarity between faces and gender difference is measured using a number of alternative geodesic, Euclidean and cosine distance between points on the manifold. In a recognition experiment we compare the perfomance distance with Euclidean, cosine and geodesic distance associated with the shape manifold. Here we explore if the distances used to distinguish gender and recognise the same for under different expressions.