Video Tamper Detection
The credibility of security devices is an important consideration in many applications. Most systems compromise two distinct parts: the core of the system, which is located in a secure area, and the edge devices which are often accessible to a greater number of people (including potential intruders and criminals). For many years tamper protection has been a standard security feature, in virtually every intruder detection device, along with external access control devices such as readers and call points from credible manufacturers. However, it is often claimed that video surveillance tamper protection is underused in security systems.
In a recent manufacturers’ forum, one prominent camera manufacturer bemoaned the fact the few installers and integrators implemented video tamper protection features. The complaint came as an answer to questions about why significant security features are often marginalised for gimmicks and on-trend functions that are used by very few applications.
Box ticking always has – and always will be – a part of the competitive process between manufacturers.
If one manufacturer introduces a function, no matter how niche it might be, there is something of a race for their competitors to include it as well. The result is that on occasions, useful features (which do not grab the headlines or add something new) are not improved.
The reality is that tamper protection in cameras has been possible for many years. Many years ago, even the most basic video motion detection systems were capable of detecting tampering and attacks on cameras. However, in the majority of applications, the location of the cameras was – in itself – a deterrent to tampering.
Since then, the falling cost of video surveillance has seen significant growth in the use of video surveillance, and often the physical structure of sites has ruled out the placement of cameras (and other devices) in locations where tamper attempts will be obvious.
With intruder detection devices and access control products, the intention of tamper protection was simply to stop the devices being opened as this allowed access to cabling. With Grade 3 detectors designed for higher risk applications, there was also a requirement for anti-masking detection and signalling.
Video surveillance was somewhat different. The range of potential tamper attacks went beyond someone opening a housing. It was not difficult to disconnect cameras, mask or spray the lenses, defocus the units, change the field of view or physically smash the camera.
When video surveillance carried a very high cost of ownership, many systems were used for live monitoring due to limitations with regard to archiving technologies. In such applications, camera tampering was either detected by an operator, or discovered when it was too late!
As systems aimed at medium-risk applications became more common, many simply recorded video and this would only be reviewed if an incident took place. Often, with such systems, this was when any tampering events (or device failures) would be spotted. If the system was proactively checked, the only indication would be an alarm LED or a visual check of the cameras.
Limitations of technology with regard to pre-event recording, alarm management, remote signalling, coupled with high costs and a low risk meant that camera tampering was something few manufacturers or installers and integrators focused on.
The reduction in the cost of ownership of video surveillance, coupled with advances in technology, mean that tamper protection is a more attractive and beneficial function. A large number of cameras are located in positions that can be accessed, and the ability to proactively create notifications of any event make the protection on offer more credible.
The introduction of smart technologies in recent years has increased the benefits on offer from video surveillance cameras. Predominantly found in networked units (although other cameras are capitalising on the move to more advanced chipsets), smart codecs are typically associated with motion detection, zone- and line-based analytics, face detection, object tracking, etc..
This trend continues to advance as ever more powerful chipsets are introduced. Manufacturers are making use of additional processing power, with many of the leading manufacturers introducing a credible level of on-board analytics.
When considering such functionality, it is common to immediately think about the detection and tracking of violations in a viewed scene. However, given the range of available analytics many manufacturers include protection for the camera itself as a standard feature.
It is fair to say that tamper protection has been a benefit of many video surveillance cameras for some years now. However, conversations with a number of installers and integrators indicate that many do not regularly use these features. Some of this is down to the location of the cameras and the way in which the system is used. There is also some hesitancy due to concerns over the performance of the features, as well as fears of additional features impacting on other performance levels.
When tamper protection first appeared on video surveillance cameras, a few of the implementations left much to be desired. If anything, in a number of cases it felt as if the functions had been added solely to tick a box on the spec sheet!
Nuisance activations did occur, and at times could not be dialled out because the functionality had little, if any, adjustability. In many applications it is simply not realistic to eliminate global scene change, objects passing close to the camera, vibrations or temporary focus loss, especially where cameras operate in both day light and under infrared illumination for day/night operation. While many of these nuisance alarm triggers are brief in nature, it equally does not mean that the manufacturer setting default trigger times of a specific duration will be any more efficient. As with any analytics, the configuration must match the situation.
Manufacturers offer varying levels of video tamper protection, ranging from simple global scene change detection through to identification of sabotage attempts such as defocusing and even impact sensing. However, the real question is whether this functionality offers credible protection, or does it merely add to the list of features and functions as a part of specsmanship?
Tamper protection comes in a variety of variants, ranging from very simple algorithms to detect global scene change, through to analytics that can differentiate between natural darkening in a viewed scene and a loss of image contrast. While manufacturers will offer different tamper protection rules, there is a high degree of similarity in most professional cameras.
Tamper protection will sometimes include detection of device malfunctions, signalling an alarm if there are high levels of image noise, if contrast becomes very low or switching to edge recording if there is a network outage or disconnection. More often than not, if these faults occur they will more likely be caused by incorrect camera configurations, failure of peripheral devices such as illuminators or a network issue. However, whether due to these reasons of as part of a sabotage attempt, it is a benefit that the installer or integrator, or the operator, can be notified as soon as the problem arises.
It is worth noting that video loss or network disconnection alerts are typically signalled from the NVR or VMS. Even where edge recording support is available, these conditions should be configured to present a notification.
Most cameras will include detection of masking or blocking, defocussing and repositioning. Some will include shock sensors to report impacts and physical attacks on the camera itself.
When testing cameras, Benchmark always looks at how the integral security features work, and whilst many have improved since the earliest implementations, it is still common for the scale of adjustability for tamper detection to be somewhat limited.
With regard to masking, most cameras offer detection of such attempts. However, in-depth testing has shown that approximately 70 per cent of cameras do not initiate a masking alarm if a transparent material is used. Such an approach does take the edge off sharpness and can be the difference between an evidential quality image for identification of an individual and a more general monitoring quality image.
Also, some cameras will not detect partial masking. The degree to which masking is missed varies by manufacturer (and sometimes by model) but some cameras will not generate a tamper alarm if 50 per cent or less of the image is blocked.
When considering detection of masking and blocking attempts, it is important that an adjustable time window for blocking incidents can be established. If a camera uses a default time with no adjustability, it probably won’t be suitable for some applications.
Good quality professional cameras will often allow the creation of a time window measures either in seconds or minutes. This enables installers and integrators to configure individual units for specific locations.
Global scene change can be used to detect if a camera has been repositioned. Tests have shown that some cameras will respond to small degrees of positional change while others require a significant degree of change for an alert to be initiated. Often the only way to identify which device is best suited to a given site’s needs is via a field trial.
Some cameras allow the capture and storage of a reference image. This can be a useful feature, expecially where an operator might not be familiar with the exact positioning of cameras. This enables them to call up the reference image and compare it to the scene the camera is actually viewing. It can also be a good aide-memoire for installers and integrators if they are called out to correct a realigned device.
Detection of defocusing is typically either selected or deselected. However, it is important that there is a time window that ensures any temporary blur that might occur when switching from day to night mode (or vice versa) occurs. Whilst best practice is to always make use of IR-corrected lenses, legacy devices might not be equipped with such lenses, and as a result some end users who are unwilling to upgrade often intend to tweak focus when required. In such cases at least the defocus detection can be used to enforce such changes. If they get fed up with the alerts it’s a good reminded of why an investment in a corrected lens makes sense!
Professional cameras allow actions to be linked with tamper detection alerts. These can include push notifications via email or SMS, visual alerts for operators, automation of recording and/or other functions, etc..
A number of cameras include integral rules which can be fully customised, as do most VMS solutions; these allow structured notifications based on scheduling too. This enables the system to send an SMS or email to a manager during working hours, allowing immediate action. Out of working hours, the notification can still be sent, but the system might also boost recording rates and image quality on neighbouring cameras or trigger a relay output on the camera.
Tamper detection alerts can be used as triggers in rules, just as other IVA alarms such as line cross or zone intrusion can. Where shock sensors are included on a camera and the VMS has full integration with the unit, the detector output can also be used when writing tamper-based rules.
Tamper detection in cameras has advanced significantly, and the available options and associated actions make sense and add benefits for many applications. Advanced communications allow real-time notifications, and that adds value for the end user.