AST May 2018 Magazine Issue Volume 23 | Page 19

There are several benefits to Volume 23 using the site’s surveillance system for intrusion detec- tion: May 2018 Edition • The same cameras can be used for detection and assessment capabilities • The per-video stream analytic cost is very low, and • Detection is not limited to just the fence (i.e. the systems can detect and potentially track in- dividuals both inside and out- side the perimeter) However, if the implementa- A perimeter protected by a fence-mounted sensor. People approaching the fence are tracked via a video analytic, enabling PTZ cameras to capture high resolution video of any tion of video analytics is to be intrusion attempt. successful (that is, provide a high probability of detection These practical considerations do not diminish the while minimizing nuisance alarms), the following power of video analytics. rule must be observed: video analytics work only as well as the quality and coverage of their video On the contrary, when properly deployed and inte- grated, video analytics offer an exciting new set of streams. capabilities that greatly enhance perimeter security at relatively low cost: This means that: • For perimeter intrusion detection applications, fixed cameras must have line of sight visibility of the entire fence line. • Cameras must be installed in such a way to avoid movement and vibrations caused by wind. • The cameras must be able to provide clear images un- der low light situations. Weather conditions including snow, rain and dense fog will affect visibility and detec- tion capabilities. • Video analytics typically require buffer space around the detection area that may be impractical in busy environments, such as urban areas with adjacent side- walks. • The algorithms must be optimized for outdoor use and be able to distinguish between people/objects of inter- est and background noise like shadows or vegetation movement. • Supplement fence sensors by providing additional de- tection capabilities (especially important for low fenc- es, ornamental fences, and walls) • Determine the direction of intrusions (ingress or egress) • Provide covert detection (no visible equipment on fence) • Detect and track people near both sides of the perime- ter fences to provide early warning of potential security events before they can occur. • This early warning can be used to direct PTZ cameras, so that high-resolution video can be captured of the in- truder at the time the alarm (from the fence sensor) is generated • Use anti-loitering analytics to detect persons “camped out” near the perimeter 17