Absolute Head Pose Estimation From Overhead Wide-Angle Cameras

  • Authors:
  • Ying-Li Tian;Lisa Brown;Jonathan Connell;Sharat Pankanti;Arun Hampapur;Andrew Senior;Ruud Bolle

  • Affiliations:
  • -;-;-;-;-;-;-

  • Venue:
  • AMFG '03 Proceedings of the IEEE International Workshop on Analysis and Modeling of Faces and Gestures
  • Year:
  • 2003

Quantified Score

Hi-index 0.00

Visualization

Abstract

Most surveillance cameras have a wide-angle field of view andare situated unobtrusively at overhead positions. For this typeof application, head pose estimation is very challengingbecause of the limitations of the quality and resolution of theincoming data. In addition, even though the absolute head poseis constant, the head pose in camera view changes dependingupon the location of head with respect the camera. In thispaper, we present a solution to estimate absolute coarse headpose for wide-angle overhead cameras by integrating 3D headposition and pose information. The work involves image-basedlearning, pose correction based on 3D position, and real-timemulti-camera integration of low-resolution imagery. Thesystem can be applied to an active face catalogger to obtainthe best view of the face for surveillance, to customerrelationship management to record behavior in retail stores orto virtual reality as an input device.