Design and Implementation of Intelligent Video Surveillance System Using Dual Cameras

Abstract

Existing CCTV video surveillance systems have difficulty in accurately identifying tracked objects. In a pan–tilt–zoom (PTZ) video surveillance system for accurate object identification, the monitoring region has a blind spot because it only tracks a specific object. Multiple PTZ cameras can be used for removing a blind spot, but this increases the delay of camera operation due to increased computational complexity. In this paper, we propose a system to remove a blind spot occurring in an existing surveillance system, using wide-area surveillance cameras and PTZ cameras. In wide-area cameras, a particle filter is used for identifying an object. Further, the moving direction of an object is predicted using motion templates, and the delay time of camera operation is reduced by controlling the PTZ cameras efficiently.