Automatic landmark detection for high resolution non-rigid 3D faces based on geometric information
Jian Liu, Quan Zhang, Chaojing Tang · 2015
3D facial landmarks play a significant role in many 3D facial studies. Due to the complex geometry of high resolution 3D human face, landmark detection remains to be a challenge problem. This paper proposes a novel method to detect landmarks automatically for high resolution 3D faces without learning or training, solely using geometric information. For high resolution 3D faces, geodesic remeshing is used to reduce the vertices number, then the remeshed 3D faces are parameterized and interpolated on a regular grid, which enable us to compute the differential geometric features more efficiently and accurately. To detect landmarks, we extract global and local constraints for each landmark by considering relative positions of landmarks and differential geometric features, then landmarks are detected by combine these constraints. We evaluate our method and compare it with other methods which are mostly related to ours in a large scale publicly available 3D face data set, BJUT-3D face database. The experimental results show that our method is robust and effective, and achieves better performance than existing methods.