Institute of Electrical & Electronic Engineers
Tracking pedestrians as they move through a camera network with non-overlapping field of view is a challenging issue due to the significant changes of their appearance among the cameras. The varying appearance of a pedestrian is the consequence of illumination changes, parameters of camera, viewing angle, and deformable geometry of people. In addition the observations of individuals are often widely separated in time and space in such systems and common proximity techniques cannot be used to constrain possible correspondences. In this paper a new feature is proposed to represent the appearance of people, which is capable of dealing with the typical illumination changes occurring in indoor environment.