Left: HDR image; right: normal image
IP cameras just keep getting better and better. High-Definition and H.264 technologies have taken security and monitoring solutions to the next level, providing clear, highly detailed images (check out the Axis P1347 and Panasonic WV-SP306 for two great examples). However, just like other cameras, there is still room for improvement. For instance, one of the key elements in getting a good image is how well a camera copes with the dynamic range in its field of view. Even the best video cameras on the market find this challenging, though it looks like new developments in High Dynamic Range video could be about to change all that. But what exactly is dynamic range and how can HDR help?
Dynamic range is the difference in brightness between the darkest and lightest part of an image. In order to capture an image with a wide dynamic range, conventional video cameras employ a variety of processing techniques. However, even the best camera will eventually struggle to cope with extreme changes in brightness. In fact, the best surveillance tool in the world for coping with a wide dynamic range is the human eye – it can adjust faster to sudden changes in lighting than any conventional camera could hope to achieve. In order to get around this, modern still-image cameras take a number of images in quick succession at different exposures. These can then be processed together by the camera or edited together on a computer to create a balanced image. This High Dynamic Range (HDR) technology is relatively easy to employ in a still-image camera, as only a single static image needs to be composited. Even the iPhone’s camera now sports a HDR function. The real challenge comes with attempting to perform this multiple exposure compositing function 30 times a second to create HDR video.
Researchers at Warwick University have recently taken on this challenge, assembling the first full-motion High Dynamic Range video system. Various attempts have been made to experiment with HDR video (including some innovative work with Canon cameras by a San Francisco-based studio) but the team at Warwick are the first to create an actual, functioning camera designed for real-world applications. Led by Professor Alan Chalmers, the team has succeeded in creating a camera that can record at 20 f-stops. The f-stop dictates the amount of light entering the lens of the camera, much like the pupil expands and contracts to allow light into the human eye. By recording at 20 f-stops, the team is able to record the same amount of exposure information that the eye sees. In other words, the camera can cope with the same dynamic range as humans. On top of this, the camera is able to capture these images at HD 1080p resolution, maintaining the clarity and detail we’ve become used to.
Capturing this amount of information leads to its own issues. Using traditional capturing and compression techniques, a HDR video camera would require 48 GB (yes, that’s gigabytes) of storage space for every minute (yes, that’s a minute) of video captured. To put that in perspective, the largest commercially available hard drive could store less than an hour of HDR video. To combat this, the team at Warwick are collaborating with a HDR technology firm to reduce these requirements by a factor of 100 through the development of new software. They have also created a HDR monitor able to display the entire dynamic range of the image. This has involved some clever sandwiching of LCD and LED technologies to provide the required brightness.
But what does this mean for IP video surveillance and monitoring? At the moment, even the best cameras are reliant on a number of processing techniques to create a balanced image. Camera manufacturers such as Axis, Panasonic and Sony use a variety of image optimisation tools to improve dynamic range. None of these are actual, real-world representations of what the human eye can see – they are approximations, albeit good ones, and they each have their limitations. A camera that employed HDR video technology would be like looking through your own eyes, able to cope with even the harshest changes in lighting. A conventional IP security camera has to compensate for sudden changes in lighting, such as a door opening, or a light being switched on and off. A HDR video camera would simply see the door opening, or the light being switched on and off, without the rest of the picture being affected. Such technology would make it much easier to track people and objects, identify faces and maintain a solid, reliable surveillance presence in the most difficult lighting conditions.
This is exciting technology. Importantly, the team at Warwick is developing it as an end-to-end solution, from the camera through compression to display. With this kind of holistic approach, technology can move forward quickly. From our perspective, IP video is ideally positioned to take advantage of these advances. High Dynamic Range is desirable in nearly every IP surveillance application, from the home user to corporate deployment. It may take a number of years for the HDR system to become a commercially viable proposition, but once it does we predict it might be The Next Big Thing. And with big advances already made in video quality and compression for IP cameras, it could be the element that turns excellent images into truly exceptional ones. We can’t wait to see more.
Image by Richard Huber (Own work) [CC-BY-SA-3.0 (www.creativecommons.org/licenses/by-sa/3.0) or GFDL (http://www.gnu.org/copyleft/fdl.html)], via Wikimedia Commons