Perimeter Protection: Analytics
When considering the protection of the perimeter of a site, there are a host of detection tools that can deliver an early warning of intrusion. These are often effective and low-cost options when the available performance is considered. However, to further enhance the level of security on offer, a growing number of end users are opting to implement video analytics as a perimeter protection tool. This leaves one final decision to be made: whether video-based or thermal imaging-based analytics offer the best degree of protection.
Previously Benchmark considered how the perimeter represents a very important element in any security plan, especially with regard to the earliest possible detection of intruders. Representing the first potential demarkation of a protected area, it defines the point at which a member of the public becomes a risk to the business or organisation.
For many, perimeter protection begins and ends with a physical barrier. A fence, wall or other boundary such as a hedge or shrubbery is established, and with many law-abiding citizens this will be sufficient to prevent unintentional access. Also, a more secure physical barrier – a fence topped with barbed wire, a high wall, heavy-duty gates, etc. – will often be sufficient to deter opportunistic criminals.
Of course, any physical barrier can be defeated if an intruder has the will and the time. Additionally, an attack on a physical element will often not be obvious to those at the site until either a breach is committed or damage is discovered after the event. As such, a detection technology is essential.
There are a wide range of stable and effective detection systems designed specifically for perimeter protection. These use varied methodologies, but it is true to say that the right devices in an appropriate environment will deliver the required results. This allows the end user to be made aware that an intrusion is taking place, thus enabling them to take appropriate action, hopefully before the intruder can reach any assets or affect an entry to buildings.
Professional detection devices fulfill their role well, but there is typically a limitation in the level of information that they can provide to a system or operator. For example, a long range external PIR covering a fence line will simply signal that activity has been sensed within its field of view; there’s no additional information provided.
Some advanced units might filter this information if active zoning is used, but in most applications the area of activity will be within the detection zone! In perimeter applications, this could be a large area. Additionally, it is not possible to tell what caused the alarm condition.
To be more proactive, an increasing number of end users are turning to video analytics for perimeter protection. However, when selecting IVA for such tasks, which options work best for effective analysis?
As with any technologies, success depends very much upon the end user having realistic expectations for the solution they are sold. This is very true with regards to IVA (intelligent video analytics). Whilst video analytics in all of its forms has some excellent use cases, deployments at the perimeter do raise a number of considerations.
Analytics work best when the violations being detected can be clearly defined. This is why IVA is such a powerful tool in business intelligence applications of when used looking for single breaches, such as breaches to traffic flow or entering specific areas. In perimeter applications, the technology is generally looking out for the unexpected, and as such the potential violations are something of a grey area!
It is also true that video analytics work best in controlled environments. Consistent lighting, limited or no activity and minimal interference (either with regards to image quality or environmental conditions) all help to deliver optimised performance for video analytics applications. Perimeter applications rarely enjoy such consistency!
These two slight negatives don’t mean that video analytics aren’t a good choice in perimeter protection. Indeed, many analytics engines have been specifically created for such scenarios. Of course, it is imperative that an understanding of what can be achieved is communicated to the end user.
The rise in prominence of video analytics has seen some very smart solutions being made available, and often these become the more marketed options. After all, if a manufacturer can do something clever, you’d expect them to shout about it! Of course, this also acts to set a standard of expectation for end users. Given the nature of perimeter applications, it is more likely that you’ll be configuring the analytics for less ‘smart’ usage.
It is vital that end users understand the need for perimeter-based analytics to detect as quickly as possible, rather than perform the advanced analysis which some system are capable of. When facing the unexpected and needing an early warning, it’s important that the analytics have a more basic role. The rules deployed at perimeters are invariably motion detection with discriminations (size, shape, direction, etc.) or line cross violations.
There are also design approaches which help counter issues with the environment, and one of these is the use of a sterile zone. This is a proven approach, and the majority of iLids recognised video analytics systems specifically approved for perimeter protection applications use a sterile zone for enhanced performance.
The concept of a sterile zone is simple. It consists of obvious demarcations – usually an outer perimeter and an inner perimeter – which people are clearly excluded from. Sterile zones don’t always physically restrict access to an area, although many do just that. The idea is to ensure that anyone entering the zone has done so with intent rather than accidentally.
As any activity in the sterile zone represents a violation, this simplifies the creation of analytics rules, which in turn optimises the performance of the analytics engine.
Carefully considered analytics rules and the use of a sterile zone make video analytics a very benefical tool when creating perimeter protection solutions. This leaves one last decision: whether to use video or thermal imaging as the source.
Thermal or video?
Perimeter protection systems invariably have to be operational 24 hours a day. In some smaller sites, their use might not be as important during normal operating hours; the hours of darkness are higher risk periods when the system must be effective.
It’s a subject that some might try to debate, but any video system requires a certain level of light to operate. It’s a matter of physics!
It is possible, with some cameras and in certain applications, to rely on ambient light for surveillance around the clock. Such circumstances are admittedly rare, and it’s more usual for surveillance-specific illumination to be required. On lengthy perimeters, this isn’t always available, and where it isn’t it can be financially unsustainable to install it. That said, in some smaller applications where lighting forms part of the perimeter solution, the provision of adequate lighting won’t be an issue.
Before assuming that thermal imaging is a better solution in applications where lighting isn’t present, it’s important to consider the pros and cons. It’s true that additional lighting – whether white light or infrared – will increase costs, but so does switching from video cameras to thermal imagers.
If light pollution isn’t an issue, white light not only allows the use of colour (always a bonus when it comes to evidential video) but also provides a deterrent effect by illuminating on-site activity for passers-by to see. Also, in the winter months when the site might be operational after dark, additional lighting provides health and safety benefits and makes the site more welcoming.
Infrared lighting can also offer a solution, minimising on light pollution, but the additional benefits (colour video and site illumination) are not achieved.
Video analytics will struggle to perform consistently if the video is noisy or has low contrast. The higher the quality, the better the IVA algorithms can work. Consistency needs to be considered throughout the full range of the protected area. For example, when protecting a building, it’s not so important if light fades at longer ranges, because intruders need to approach the building to affect an entry. With perimeter protection, illumination must be consistent along the whole perimeter line.
A final point about infrared light is that it can throw up some unusual effects in certain conditions. Whilst these are easy to spot and understand when viewed by a human operator, an IVA engine will typically see them as change within the scene.
Thermal imaging views objects based on radiation, and as such does not require lighting at all. Also, because of the use of colour pallettes such as ‘white hot’, any intruder will stand out with some degree of clarity. For an analytics engine, this makes detection of violations a simpler and more efficient process.
Thermal imagers usually have much lower resolutions than video cameras. Despite this, performance with well-implemented video analytics such as motion detection or line cross violations is often very good, and in difficult conditions will out-perform video alternatives.
There are an increasing number of low cost thermal imagers with very low resolutions, and with these effective range can be limited. Forget quoted operational ranges as these are based on the Johnson Criteria which is not security-based. However, in Benchmark tests we have seen IVA being effective on such devices even where it was challenging to identify targets by eye. Of course, there is a risk of nuisance activations when the resolution is very low.
The downside of thermal imaging is that it operates as a detector, albeit one with a motion-based image to provide a greater degree of information. The footage will never provide identification or deliver usable evidential information to the degree that video can.
Video analytics is a very useful tool for perimeter protection, and can deliver a credible level of security. Video and thermal imaging have differing pros and cons, which is why many systems end up deploying both!