A biologically motivated system for tracking moving people using a stationary camera.

摘要

This thesis describes the design and implementation of a real-time system for tracking moving people in an indoor environment. The system utilizes a stationary camera which can be mounted in any location within the indoor environment such as on a wall or the ceiling. Embodied in the system is a control mechanism which is based on a biological model of gaze control. This mechanism controls the movements of software windows which are used for tracking, such that the target is always centred within a window, in much the same way as human eyes move such that a target is always centred on the foveal region of the retina.;Results from tracking experiments show that the implemented system is capable of tracking both single and multiple targets in real-time, with the camera placed at different heights above the floor. Furthermore, the system performs just as well under different lighting conditions.;The tracking system requires no special hardware for its operation. All that is required is an off-the-shelf processor, camera, and frame-grabber. Therefore, the system provides a cost-effective solution for many applications which involve tracking such as security surveillance, robot navigation, and (human) traffic analysis.