Samsung patents video surveillance software that could help Big Brother spy on you more effectively

We live in an era of intelligent machines. Smart lights. Smart thermometers. Smart watches. The notable exception are cameras, mainly because they usually come attached to other smart devices. But the gap is now finally closing with the help of Samsung, although that may not be necessarily for the better.

One of the latest patent filings from its defense subsidiary reveals a piece of homegrown software that could help standalone recording systems – particularly those used for surveillance – become intelligent in their own right. At the center of the invention is a low-pass filtering algorithm that makes clever use of algebra to single out regions where activity is detected.

samsung_1

The pixel values of the inactive areas are multiplied by a predetermined coefficient until surpassing a cutoff point past which the data is no longer registered in the backend system to which the camera is connected. The busier a section, the lower the coefficient set by the software, which means that the camera is at its most sensitive in the low-activity regions that intruders are the likest to target.

But that sensitivity optimization doesn’t come into play in the second and possibly main use of the software that Samsung lists in its filing, which is to single out specific people – or targets, as they are referred to in the document – when they enter a camera’s view. In practice, a separate facial recognition function would handle the actual identification in that scenario, but the low-pass filter serves a no less important role: conserving bandwidth.

[do_widget id=text-11]

Pulling a feed from an anti-theft camera above a garage door, or even several dozen cameras in a closed circuit around a building, doesn’t tax the network too much. But when you have upwards of thousands of cameras streaming many terabytes of real-time video each and every second over large distances, filtering out unnecessary data can help save a lot of valuable time on the journey to the backend.

samsung_2

And the less delay there is in the video feed, the faster the operator can react. That’s a key factor in crime response, but Samsung’s technology could potentially also find use in much more questionable applications that may not appeal all that much to consumers who thus far mostly benefited from the smart appliance trend.

The fact that the rights for the invention belong to its defense subsidiary means that it will almost certainly be targeted towards the public sector, where the use of surveillance has attracted a great deal of alarm in recent years. The prospect that government agencies may soon be able to find their targets even faster and more efficiently than now with Samsung’s help is concerning to say the least.

[do_widget id=text-6]