Reframing security: Video analytics are changing situational awareness and forensic search
We live in a world of lenses. Video security surveillance is everywhere; Security News Desk estimates that there are more than 245 million surveillance cameras in action worldwide. Ideally, the goal of video surveillance is to improve public safety and crime-fighting tools. However, a key challenge when it comes to all these lenses is putting the footage to effective use.
Video analytics have the ability to solve that problem. Powerful software solutions are creating more reliable and interlinked approaches to scanning, identifying, interpreting and acting on the information that surveillance video provides.
Thanks to evolving analytics, video situational awareness and forensic image search techniques are coming into new and sharper focus across the security-surveillance landscape.
Fewer false alarms
When cameras are hooked up to analytics software that can identify objects, security systems will create fewer false alarms. A primary advantage of video analytics is that modern software can identify collections of pixels as actual objects, rather than simply noting changes in pixel quantity within a shot. That means whether it's a car, a person, a deer or just a swaying tree branch, the analytics at work can distinguish between types of things by categorizing images from a library of known shapes. The major selling point of this advance is that surveillance officials receive fewer false alarms for the kinds of objects that aren't of interest, such as wildlife and windblown trees, allowing them to focus their vigilance on high-priority objects such as vehicles and people.
Better search capabilities
Object identification also translates into more powerful forensic search. Analytics that identify particular shapes create a significant tool for post-event review. With a library of shapes stored in the forensic image investigator's database, video experts can now query hours of footage for the exact arrangements of pixels that match the shapes they want to see.
With video analytics, multi-camera search can move beyond facial recognition. Searching for a single suspect has historically revolved around facial-recognition software hooked up to a network of surveillance cameras. That means searchers counted on a camera capturing the suspect's face in a fairly direct fashion for the image to register a match, adding the need for hours or even days of search time, not to mention some good luck.
As technology has evolved, video analytics hooked up to camera networks can now make matches based on a suspect's full body and track a person's movement based on recorded shapes and movements. The Washington Post explained that this type of intelligent surveillance can help law enforcement track criminals through malls or neighborhoods. The technology is also used to provide enhanced security at major public events, such as political rallies or big sporting events.
Video analytics have traditionally relied on algorithms and rules-based systems to provide situational and forensic tools for security experts. The future, however, is about even smarter cameras and networks — ones increasingly capable of machine learning. For instance, government researchers are currently working on artificial visual intelligence systems, according to Forbes. Under this model, video surveillance tools can predict potential threats and hazards based on how objects move and behave within a shot. The potential to give public safety and security officials that kind of warning means that in-the-moment monitoring could achieve ahead-of-threat responsiveness.
Because the lenses that watch over us while we work and play are there to protect us and help apprehend individuals who have harmed others, it's crucial that the images those cameras captured can be analyzed at the highest level possible. We want our surveillance to be dynamic, discerning and capable of capturing the critical data we need in real-time. Video analytics are the key to making our systems work in just that way, now and in the future.