Robust 3D Head Tracking Using Camera Pose Estimation

In this report we present a robust method to recover 3D position and orientation (pose) of a moving head using a single stationary camera. Head pose is recovered by formulating the problem as a camera pose estimation problem. 3D feature points (artificial or natural occurring) are acquired from the head prior to tracking and used as a model. Pose is estimated by solving a robust version of "Perspective n Point" problem. The proposed algorithm can handle self occlusions, outliers and recover from tracking failures. Results were validated by simulations and by a comparison to an accurate magnetic field 3D measuring device. Our contribution is a system that is not restricted to track only human heads, and is accurate enough to be used as a measuring device. To demonstrate the applicability of our method, three types of heads (human, barn owl, chameleon) were tracked in a series of biological experiments.

• Type: Conference
• Year: 2006
• Where: ICPR
• Authors: Ehud Rivlin, Shay Ohayon
• Keywords: Computer Vision, Head Pose, ICPR 2006, Measuring Device, Pose Estimation Problem