3D Head Tracking by Particle Filters
Fadi Dornaika, Franck Davoine, Mô Dang · 2004
In this paper, we propose a particle filter framework for 3D tracking of heads and faces in monocular video sequences. We propose two different approaches. The first approach utilizes a statistical facial texture model as an observation likelihood. The second approach utilizes a deterministic facial texture which is built on-line. The developed approaches have been successfully tested on several video sequences. The resulting 3D head tracker can find applications in many fields, such as human-computer interaction. 1.