Dynamic Object Path Detection in a Network of Surveillance Cameras

Abstract

Today, automated camera surveillance systems play a major role in securing public and private premises to ensure security and to reduce crime by detecting behavioral changes of moving objects. The important goal of such a surveillance system is to reduce human intervention while at the same time, provide accurate detection of moving objects. Many researchers have attempted to automate different aspects of camera surveillance such as tracking humans, traffic controlling, ground surveillance, etc. However, a system that overcomes overall difficulties that arise in the task of object detection and object tracking has not been developed because of high variance in the problem domain. The proposed system tracks the path of a locked object through a network of cameras. In contrast to traditional methods where the operators have to switch the screens manually to find the target objects, the proposed technique, once locked to an object; automatically tracks it through a camera network and generates the path on a map. We propose to use stereo cameras to enhance the detection and tracking of objects in 3D space.