Technology has come a long way since the Sept. 11 terror attacks. ID cards are a lot smarter; biometrics read eyeballs and fingerprints; sensors sniff out explosives; cameras see in the dark — and sophisticated software brings everything together.
The need for security is driven by any of several possible threats associated with the nation’s critical infrastructure (highways, malls, chemical, nuclear or oil plants, etc.), as well as those that impact the lives of millions, but technology alone will not save anyone or anything without a first-rate and professional assessment of your organization’s risk.
Risk assessments enable organizations to integrate their financial objectives with their security objectives, and to make valid security-technology choices. After all, these brilliant devices do not come cheap. If you and your security team do not know what threats you face, then the core building block of your entire security design may be flawed. You will not know how to, or be able to, adequately protect your assets, including your employees, equipment, intellectual property and even your brick-and-mortar buildings.
In addition, if a professional consultant or your in-house security experts are not used to assess your security needs prior to designing a system, the risk of incurring liabilities stemming from the improper implementation of an unstructured program could arise. You risk the loss of efficiency if, for example, interoperability is not considered by the design team. More importantly, you could be protecting the wrong assets or addressing the lesser vulnerabilities while missing the critical vulnerabilities.
Security Flaws on the Perimeter
Traditionally, perimeter security has relied on humans staring at video scenes coming in from CCTV cameras set along the perimeter. Some of these systems can comprise hundreds of cameras and dozens of head-end monitors — and with such a high flow rate of image data, it is not prudent to depend solely on a human operator to monitor events of security significance. The cameras may or may not be coupled with fence sensors and radars, but either way, the security crew ends up monitoring several screens to visually detect or confirm the intrusion. The concentration span of a human watching video monitors and attempting to detect intrusions has been proven to be in the neighborhood of less than 30 minutes. So, a skilled intruder can defeat both the system and the human monitor in many ways.
Even if the system is properly maintained, data coming from a security fence can only provide information about the approximate location of the intrusion. Once the perimeter is breached, security personnel may lose the intruder unless he or she is manually followed with a pan/tilt/zoom camera. Having enough time to complete this task, notify first responders and intercept them — all before the adversary is able to complete their mission — is therefore a gigantic challenge.
Knowing where the threat originated, where it is now, and where it is going is the key objective of any effective security system. Traditional perimeter sensors give very little real-time information to responders. For example, a fence motion sensor surrounding a critical facility with an 8,000-meter perimeter notionally has knowledge of the 8,000-meter fence line. But the fence sensor does not provide discrimination in terms of object recognition, and it only alarms when the fence shakes. It is not enough to be notified of a potential breach at the fence-line perimeter. Intruders must be spotted, identified and tracked from outside the fence; and in a world comprised of the uncontrollable — changing light, clouds drifting, wind blowing, snow falling, dense fog, darkness, etc. — the probability of detection is severely complicated by high false-alarm rates.
“Machine vision technology,” which can spot static problems such as a defect in a semi-conductor chip, has been around for a long time, but historically, it did not cope with an uncontrolled environment. It has to look at a template to spot a problem. Today’s threats have no templates. They demand the most sophisticated image analysis technologies available, and that includes the ability to put real-time detection into a geographic context.
Video Image Analysis for Perimeter Protection
Once you have identified the vulnerabilities your company or facility faces, a system design that addresses the vulnerabilities is crucial. A key issue facing security managers today is the scalability of the security system. Not only does the system need to scale to address one or more facilities, it needs to do it in a way that does not require significant increases in staff. This is where machine vision automation can really help.
Thanks to vast progress in computer-processing horsepower, multiple CPUs can now be put into one computer, which allows for a significant amount of image analysis to take place — up to 16 times more than was available just a couple of years ago. This technological leap now makes it economically feasible to apply advanced image processing techniques for intrusion detection, recognition and tracking that only federal armed forces could afford less then a decade ago.
Video image processing or “video analytics” is an advanced tool used in some security systems to automatically analyze video feed for a number of characteristics and patterns. It combines a number of inputs from cameras, radars and fence-detection systems, automatically detecting, evaluating, ranking and reporting potential threats. This automation alleviates the security guard of an impossible (and unscalable) task of manually monitoring hundreds of cameras across many kilometers of fence line. Systems like this can help automate the process of threat assessment. A legitimate threat is one that has the capacity and intent to do harm.
Video analytic systems detect moving objects of significance while ignoring motion due to changes in light and weather. These systems also classify objects into a variety of groups, including people, cars, trucks, etc., and they are now combined with behavioral analytics, which go to the intent of the intruder by noting deviations from expected behavior. This provides a very effective way to monitor for specific types of events, and has great value in terms of reaction, giving your security team a chance to be proactive rather than simply reactive.
In addition to visible-light video streams, video analytics can typically detect and track objects using infrared and thermal camera video streams. These cameras can see in the dark without the use of lights, and can to some extent see through fog, which other cameras cannot.
But perhaps the biggest breakthrough in beyond-the-fence protection comes from the combination of video analytics with a Geological Information System (GIS). This design element helps security personnel to immediately assess how close the threat is, how fast it is moving and in what direction it is going. The military has been using this type of technology for many years, and cities and states have used GIS for surveying, but now it is economically feasible for use in security.
Thanks to the ability to locate objects and people geographically, the size and mass of the intruder may be determined. Something or someone approaching a camera straight on does not look like it is moving, and since most systems rely solely on motion, an intruder could easily slow or escape detection. Only a few companies are pursuing methods of locating both moving and non-moving objects, which increases the probability of detecting an approaching target at significant distances. Additionally, the ability to geo-locate threats is rare: virtually 90 percent of vendors do not have such capabilities.
If properly considered, a positioned, integrated, sensor system — which includes ground surveillance radar, video analytics and geo-spatial abilities — gives the security professional a viable chance to interdict and respond to a threat.
Larry Bowe is general manager of ICx Vision Systems, a business unit of ICx Technologies.