Intelligent people surveillance with a cooperative camera network

This paper presents a novel intelligent surveillance system using a network of cooperative cameras which can be utilized to detect multiple people in the scene and identify them with unique labels. Camera calibration, homography constraints and field-of-view lines are integrated and used to estimate object locations on a reference plane which is the ground plane as seen from a virtual top-view. Hence, locations of people under surveillance can be displayed to a system operator in real time. The experimental results further demonstrate the efficiency of the proposed system.