3D face tracking using appearance registration and robust iterative closest point algorithm

Recently, researchers proposed deterministic and statistical appearance-based 3D face tracking methods that can successfully tackle the image variability and drift problems. However, appearance-based methods dedicated to 3D face tracking may suffer from inaccuracies since they are not very sensitive to out-of-plane motion variations. On the other hand, the use of dense 3D facial data provided by a stereo rig or a range sensor can provide very accurate 3D face motions/poses. However, this paradigm requires either an accurate facial feature extraction or a computationally expensive registration technique (e.g., the Iterative Closest Point algorithm). In this paper, we propose a 3D face tracker that is based on appearance registration and on a fast variant of a robust Iterative Closest Point algorithm. The resulting 3D face tracker combines the advantages of both appearance-based trackers and 3D data-based trackers. Experiments on real video data show the feasibility and usefulness of the proposed approach.