3D head tracking using normal flow constraints in a vehicle environment

Abstract

Head tracking is a key component in applications such as human computer interaction, person monitoring, driver monitoring, video conferencing, and object-based compression. The motion of a driver’s head can tell us a lot about his/her mental state; e.g. whether he/she is drowsy, alert, aggressive, comfortable, tense, distracted, etc. This paper reviews an optical flow based method to track the head pose, both orientation and position, of a person and presents results from real world data recorded in a car environment.