CCTV video analytics: Recent advances and limitations

Abstract

There has been a significant increase in the number of CCTV cameras in public and private places worldwide. The cost of monitoring these cameras manually and of reviewing recorded video is prohibitive and therefore manual systems tend to be used mainly reactively with only a small fraction of the cameras being monitored at any given time. There is a need to automate at least simple observation tasks through computer vision, a functionality that has become known popularly as "video analytics". The large size of CCTV systems and the requirement of high detection rates and low false alarms are major challenges. This paper illustrates some of the recent efforts reported in the literature, highlighting advances and pointing out important limitations. © 2009 Springer-Verlag.