Human recognition technologies has gained a lot of attention during last years. Many different methods have been developed based on fingerprint, eye iris, voice and face.

Face recognition technologies become one of the most popular human recognition approaches since face identification system does not require any advanced hardware and physical contact.

Face recognition approaches could be categorized as: face recognition based on images, face recognition based on range scans or point clouds, multi-modal 2D and 3D face recognition.



Most of efforts have been devoted to face recognition from two-dimensional images. They can achieve good performance but they are still sensitive for facial expressions, head position and lighting conditions. To overcome these limitations, researchers focus on three-dimensional approaches; 3D models are insensitive to the illumination and pose changes.



In our works we focus on 3D face pre-processing. The purpose of pre-processing is to minimize the influence of quality to the recognition step. Three-dimensional scans contain artefacts like holes and spike; to gain good results at the recognition step we have to deal with them. Then, automatically anthropometrics points localising gives us a better comprehension of the 3D face model in terms of validation, orientation, etc. and constitutes an essential step for 3D facial recognition.