Securing Outdoor Assets with Trusted Alerts

Safeguarding outdoor assets in a reliable and cost-effective manner often comes down to a single requirement – accurate intruder alerts and timely information about the unfolding event. While there are many technologies available for outdoor security, smart cameras with video analytics have emerged as the solution of choice for detecting intruders in real time outdoors.

Yet the best technology will be handicapped if the alerts generated cannot be trusted. Repeated false alarms can eventually condition security operators to ignore real intrusions, undermining trust in the perimeter security system.

In most cases the shortsighted response is to single out the security force as scapegoats, which ignores the real problem – alert fatigue. After responding to hundreds of perimeter breach alarms that turn out to be nothing more than small animals or windblown branches, even the most conscientious security guards lose confidence in the system and start to ignore its warnings.

There is no longer any reason for this situation to exist. This design guide relates how smart video security technology, when properly deployed according to best practices, can cost-effectively protect outdoor assets with high accuracy and low nuisance alerts to help security forces stop intruders before they act.

Start with the best detection: Use smart thermal cameras

Viable outdoor security must start with a sensing system that is accurate, 24-hours per day. For this reason, conventional wisdom asserts that smart thermal cameras are the best system for detecting intruders outdoors. This is because thermal cameras see heat rather than light, so they are a perfect ‘human detector,’ and will ignore headlights, reflections off water, and other light-based activity, expanding their usefulness from their traditional role as night vision cameras to 24-hour intrusion detection solutions.

Smart thermal cameras with built-in video analytic software offers several advantages:

  • They detect in the dark with no need for costly artificial lighting.
  • They work 24 hours/ day.
  • They ignore reflections, shadows, moving headlights, direct sunlight, and other light-based phenomena that can trigger alarms in a visible camera detection system.
  • Because humans give off heat, thermal sensors are far more effective in spotting a person than visible cameras.
  • They detect body heat as far away as 600 meters – a third of a mile.
  • A single thermal camera can protect an area the size of a football field.
  • Proper physical design makes them immune to the effects of weather and other environmental factors.

In the past, the higher price for thermal technology limited their use in commercial applications, but as costs continue to fall, many organizations are now able to choose thermal cameras as the foundation for their outdoor detection applications.

Geo-registration and detection accuracy

Smart thermal cameras are designed to detect movement, but outdoors, everything moves. A smart camera must be able to tell the difference between small objects such as leaves or debris and a person entering a secured area. One of the best ways for a camera to make this determination is through ‘geo-registration’ which provides the actual location and true size of all pixels in the camera’s field of view.

Consider how human vision works:

Our eyes give us depth perception – we can tell which object is close and which is far. But a ‘one-eyed’ camera can’t, unless it’s geo-registered. For example, a small animal near the camera will look much larger than a man at 300 meters away. (Figure 1)

A smart camera needs to ignore the animal at right while alerting on the distant person, even though the animal will cover more of the camera’s field of view. The same approach applies to blowing trash, clouds, and other moving things which are always present outdoors. With a camera that is geo-registered, such non-security related movement will be ignored and will not send alarms.

Essentially, geo-registration enables a three-dimensional capability for a smart thermal camera. From this information, geospatial analytic rules can be used to eliminate movement based on size while still detecting human-sized intruders under all conditions.

Geo-registered analytics in action: From-to Zones

Motion zones are often used by video analytic systems to detect the movement of objects and to send an alert to notify security that an intruder has been detected. By default, any object moving within a motion zone triggers an alarm. However, when used for outdoor applications, motion zones can lead to an abundance of nuisance alerts because they lack the discriminating intelligence to recognize the difference between ‘unimportant’ movement caused by the natural environment and ‘relevant’ movement that represents a security threat.

Cameras that are geo-registered can create more intelligent rules called From-To Zones, an important tool for reducing nuisance alerts while maintaining a high probability of detection.

Targets detected in a From-To Zone will only trigger an alarm when a specifically sized object – such as a person – moves from one zone into another defined area of the camera’s field of view. Correspondingly, objects that are not detected coming from one zone into the other are ignored.

From-To Zones are a very powerful method for reducing unwarranted alarms. Importantly, they can be configured to detect zones that are geo-registered to the ground. This means From-To Zones will only alarm when a person’s feet have been in the ‘From’ and then enter the ‘To’ area, while ignoring detections that only show a part of a person such as their head. This is particularly useful when the security area includes a fence, and you only want to detect pedestrians who have crossed over the perimeter into the security zone.

To see how From-To Zones work in the real world, consider an application where you need to detect pedestrians approaching the perimeter, but are not concerned about people leaving the building.

With From-To Zones, the camera will only trigger an alert when intruders move towards the facility – ignoring everyone else, and greatly reducing unnecessary alarms.

For another example, consider a windy perimeter around an active construction site where trash blows around the scene. Inevitably, the trash will collect along the fence and grow in size until it is large enough to trigger an alarm. When geospatial From-To Zones are employed, the smart camera will ignore the individual pieces of trash that did not move from one zone to the other. It will also ignore the large collection of trash along the fence, even if it matches a size rule, because the camera will have already determined the actual size of each piece of trash as it traversed the field of view.

To take advantage of this powerful functionality, choose cameras that are geo-registered as a core feature.

Eliminate camera movement with stabilization

Many intrusion detection systems are deployed along open areas that are naturally impacted by high winds or vibrations from planes, trains, weather and machinery. Without image stabilization, these applications can be overwhelmed by nuisance alarms or worse, outright misdetects. It is difficult for smart cameras to detect movement in a scene when the whole field of view is also moving from camera shake.

For indoor surveillance applications, such camera shake is rarely a problem. In the outdoors, where cameras are mounted high on poles, even a slight wind or vibration can cause nuisance alarms.

The best way to overcome the impact from wind or vibrations is to choose smart cameras that stabilize the image electronically, before the video analytic rules are applied. Look for cameras that use electronic or gyro-based stabilizers as a foundation for their detection capabilities.

Avoid server based video analytics outdoors

If its imager is a thermal camera’s eyes, image processing is its brain. It is image processing that turns what is essentially a passive surveillance tool into an active security device, detecting motion and recognizing targets under any conditions.

Cameras which employ both a high degree of image processing and on-board video analytics have a great advantage in accuracy and detection distance over solutions that employ analytics on a server, outside of the camera.

For one, cameras without embedded analytics must compress the video data for transmission over the network. That’s a problem because compression removes most of the finer scene details – up to 99% of the original data – seriously degrading a video analytic system’s ability to accurately detect and recognize targets. In fact, on days with restricted visibility due to rain or fog, data compression has caused smart cameras to miss virtually all intruders in a scene.

On the other hand, when the uncompressed imagery is processed in the camera, 100% of the raw scene data is available for analysis. With on-board image processors examining the full visual detail of every video frame, you can achieve a much greater degree of accuracy in detecting motion and recognizing potential threats.

Integrating the imager directly with the video analytics – inside the camera enclosure, with a high degree of processing the edge – is the cornerstone of a smart camera’s ability to accurately detect targets in the outdoor environment.

Strategies to increase probability of detection

Smart video systems are a great way to protect outdoor assets, but the system has little value if people can enter a secured area undetected. Here are some pointers to avoid gaps in coverage and make sure your system detects every time.

Determine your camera’s true detection range

A perimeter security system based on video analytics operates by ‘seeing’ targets that move into a camera’s detection area; knowing the camera’s true range lets design a dependable system with no coverage gaps.

Unfortunately some manufacturers specify camera ranges that overstate their detection capabilities. This means it’s up to the integrator to determine the camera’s true detection distances; otherwise your perimeter solution may leave gaping holes that can allow intruders to pass through undetected.

The best practice to determine a camera’s true detection range is to measure the farthest distance at which the camera can automatically detect a person walking ‘inbound’ or directly toward the camera, as shown in Figure 3. Inbound detection is always less than crossfield because a person moving across the camera’s field of view creates a larger amount of motion, which is easier to detect. In contrast, a person walking toward the camera produces very little motion, making the target more difficult to detect. In the real world, intruders can enter a perimeter from any direction, so it’s important to design the system for all situations.

Adjusting for challenging weather environments

For a conservative design that allows smart camera to detect reliably in less-than ideal conditions, it’s a good idea to de-rate the distances to 80% of a smart camera’s normal inbound detection range. In areas of extreme weather and heavy fog, further de-rating may be advised.

At one time such de-rating was cost prohibitive, but now that thermal camera prices have fallen so dramatically, and there is no longer a large cost difference between longer- and shorter-range smart cameras such an approach makes good security and good business sense.

Cover the blind spot

A camera’s field of view doesn’t begin where it’s mounted. Instead it can only detect at a measurable distance in front – this is the blind spot. Every security camera has a blind spot, and this must be considered in the perimeter security design, or someone will be able to walk right under a camera undetected.

Some outdoor surveillance designs will narrow a security camera’s field of view to increase the camera’s detection distance in an effort to decrease costs. This is not necessarily a bad concept, but it’s important to understand that doing so also makes the blind spot under the camera larger, sometimes doubling the number of cameras required.

To provide complete coverage, the view of each camera must be designed to cover the adjacent camera’s blind spot, as shown in Figure 4.

Perimeter design software can help

Some manufacturers offer design tools that can help you model a camera layout using a Google map of the area under consideration. This is a good practice to check detection distances ahead of time and ensure that blind spots are properly addressed. SightLogix offers such a tool, called SightSurvey.

Conclusion

Today’s smart video is an ideal solution to the new challenges in site protection that confront security professionals. It outperforms older technologies by a wide margin. It often costs less. Installation is less disruptive, and the technology is highly reliable. Essentially, a smart video security system is a force multiplier, taking the burden of monotonous surveillance off regular security staff. Instead of just watching endless video feeds, the staff gets information that lets them do their jobs better. When deployed using best practices of product selection and installation, smart thermal video is the obvious choice for outdoor site security applications.


 

John Romanowich
President & CEO

SightLogix, Inc.


 

Leave a Reply