A robust servo based headtracker with auto-zooming in cluttered environment
Teng-Kai Kuo, Cheng‐Ming Huang, Li‐Chen Fu, Pei‐Ying Chen · 2003
In this paper, we establish a nearly real-time surveillance tracking system, which is able to detect any person who intrudes to a prohibited area and lock his head image at the scene center. The entire underlying algorithm consists of a motion detector, a modified ellipse head tracking algorithm, auto-zooming ability, and a VPDA filter which is modified by probabilistic data association filter (PDA filter). The system operates about 35 ms and actively controls the camera platform pan and tilt motion to track a person in real environment. Finally, a number of experiments are conducted to validate the effective functionality of the head tracking system.