Protecting the Perimeter
- By Eric Eaton
- Dec 01, 2009
Traditional security planning focuses on protecting the perimeter of a facility. However, escalating threats and increasingly lethal intrusion capabilities are redefining the security perimeter and require technologies that keep security forces a step ahead. Effective security interdiction must now begin with actionable intelligence -- understanding and recognizing threatening behavior before an intruder gets anywhere near the target or executes a plan.
Security video is either used for forensic analysis after an incident has occurred, or it employs a limited-capability technology known as video analytics -- a video-motion and object-classification- based software technology that attempts to watch video streams and sends an alarm on specific pre-programmed events.
This legacy solution, however, generates a great number of false alarms that effectively renders it useless in the real world.
A Smarter Solution
After witnessing instances where unmanned surveillance cameras were deficient in adding adequate security to critical environments, and consulting with sources in the security-technology community, BRS Labs created the paradigm-shifting new science of behavioral analytics, dramatically improving physical security through the application of patented, video surveillance behavioral analysis.
The technology uses cognitive reasoning, much like the human brain, to process visual data and identify criminal and terroristic activities. Built on a multi-patented framework of cognitive learning engines and computer vision, the software product, AISight™, provides the world’s first fully automated, easily scalable surveillance solution that analyzes behavioral patterns, activities and scene content without the need for human training, setup or programming.
The behavioral analytics technology analyzes video in real time, observing what happens in the scene, frame-by-frame, and learns the behavioral patterns that occur over time. This creates a platform conducive to enhanced situational awareness.
Real-time, relevant alerts. AISight isn’t limited to a set of rules. It actually learns the normal range of behavior for an environment and sends alerts when objects behave outside of those norms. This produces real, actionable alerts.
Total field of view. AISight is capable of observing and learning a scene, enabling it to understand that tree limbs moving in the wind pose no threat. No masking is required, so the CCTV field of view provides 100 percent surveillance coverage.
Adaptability. Another benefit of using AISight is its ability to adapt to changes in day-to-day operations. For example, at a vehicle entry gate for a secure facility, AISight would learn that it is normal for vehicles to move only one way, and would alert if vehicles moved in the wrong direction.
Easy installation. While non-learning video analytics products can require personnel to be onsite for days or weeks to define zones and rules, AlSight only requires the connection of the video feeds to its equipment, which can be done by a trained integrator in hours or days. After installation, it quickly and automatically learns the scene on its own and begins generating alerts after it has learned the scene. This ease of installation enables BRS Labs to bring value to a client’s security operations faster and provide feedback that helps with refinement of the total security infrastructure, including camera positioning and providing a distribution list for alerts and reporting protocols.
Behavioral Analytics
Advances in cognitive science and the understanding of human consciousness led engineers to invent hypocepts, the technique of representing “hypothetical concepts” that explain observational data.
The video analysis engine is the observation engine. Computer vision techniques, such as background subtraction, tracking and scene modeling, are used to observe the activity occurring in a scene. Each observation is encoded and passed to the learning portion of the cycle.
The machine learning engine uses the encoded observations to build and refine hypocepts that describe the observational data. These hypocepts are decayed or reinforced based on the frequency by which observations occur. Over time, as more cycles are completed, the hypocepts mature, allowing the software to learn what activity is typical and, more importantly, what activity is abnormal.
As new observations are made, the MLE compares each new observation with its library of hypocepts and color codes it. These alerts are fine-tuned by the security personnel at each installation, so each system is unique.
Customers can configure the software-response engine to automatically capture and/or dispatch alerts that correspond with each color band. Each user in the system can subscribe to one or more color bands of information. Each color band is designed to provide significantly more information than the next hotter color band. Thus, if a customer only desires to see the most unusual events, that customer would subscribe only to red alerts.
This configurable temperature selection acts like a thermostat by giving each user the ability to turn up or turn down the volume of information provided by the software.
Behavioral analytics software currently provides real-time alerts to any type of mobile PDA device from CCTV security sensors. Compliant with industry-standard HTML protocols, BRS Labs can immediately deliver actionable alerts to security personnel while seamlessly integrating into existing commandmanagement software systems.
The proposed solution provides intelligence community security operators with a beyond-the-fence awareness that promotes a proactive response. Increased information sharing enhances results for the deployment of multi-agency response personnel.
This article originally appeared in the December 2009 issue of Network-Centric Security.
About the Author
Eric Eaton is the chief technology officer at BRS Labs.