Analysis and Retrieval of 3D Facial Models Using Iso-geodesic Stripes

Abstract: In this paper, we describe a framework for analysis, representation and matching of three-dimensional faces. Basic traits of a face are encoded by extracting iso-geodesic stripes from the surface of a face model. A compact representation is then constructed through a modeling technique capable to express the basic shape of iso-geodesic stripes and quantitatively measure their spatial relationships in a three-dimensional space. This information is encoded in an attributed relational graph. In this way, the structural similarity between two face models is evaluated by matching their corresponding graphs. Experimental results on a 3D face database and baseline comparison show that the proposed solution attains high face recognition accuracy and is reasonably robust to facial expression and pose changes. Experiments also target the identification of portions of the face surface which are most significant for the purpose of discriminating between faces.