In this paper, we present a novel approach for face recognition and authentication based on dimensional surface matching. While most of existing methods use facial intensity images, a recent set of approaches focus on introducing depth information to surmount some of challenging problems such as pose, illumination, and facial expression variations. The presented matching algorithm is based on ICP (iterative closest point) which aligns one presented probe model to a 3D face model from the gallery data set and provides perfectly its posture. Recognition score is given by a region-based similarity metric which takes into account regions' labels. Here, a specific study in facial expression analysis is done for the labelling purpose. We aim by means of this study at a best way to split and segment the 3D gallery's faces. A new multi-view registered 3D face database including significant variations (pose, illumination, and particularly facial expressions), is collected in order to achieve experiments on our proposal