Face modelling and tracking from range scans

This paper presents an automatic framework for face model and texture reconstruction, facial expression transferring, and face tracking from 3D and 4D range scans. Taking the scale transformation into consideration, an extension of the iterative closest points (ICP) algorithm is introduced to solve the problem of registering a generic face model onto range scans with different sizes. A deformable model derived from thin shells is proposed to faithfully reconstruct face models and textures from range scans. Dense point correspondences are then established across different reconstructed face models so that facial expressions can be easily transferred from one face onto another by the deformation transfer technique. The proposed deformable model is also used to track the facial motion presented in 4D range scans and some preliminary results are obtained