Fredrik Nilsson is general manager of Axis Communications and an authority on IP surveillance.
In a typical DVR-based surveillance system, video intelligence has to be centralized after the point of recording.
A distributed intelligence using analog cameras puts the analysis algorithms in the video servers, which also convert the analog images into an IP stream.
Distributed intelligence systems can, of course, use network cameras that have the analytics built into the cameras themselves.
Fredrik Nilsson General Manager, Axis Communications
Network video allows for new capabilities in the surveillance industry that were not feasible in an analog environment, either because they were impossible to implement, or just too cumbersome. Some of the hottest new technologies available in a network video installation are intelligent video, megapixel cameras, and something called immersive imaging.
Today, far more video is being recorded than anyone could ever monitor or search. Studies from the Sandia National Laboratories, which develops science-based technologies to support U.S. national security, suggest that personnel can only watch one monitor for up to 20 minutes before losing focus. Without some form of built-in algorithm compiling relevant information, there is simply no way to monitor all the surveillance cameras in a system - unless you've got an almost unlimited budget.
That's where video analytics enters the picture. Intelligent video (IV), the next big trend in video surveillance, will allow cameras to monitor events within the field of view. Advanced network cameras can have built-in motion detection and event handling. In addition, more intelligent algorithms, such as automatic number plate recognition (a.k.a. license plate recognition) and people counting, are being integrated into security and surveillance systems. Network cameras and IV have important synergies that make the systems more reliable and effective than those using analog cameras with a digital video recorder (DVR) or other centralized system.
Intelligent Video Defined
Different vendors have referred to IV by various terms including "actionable intelligence", "video analytics", and "intelligent video". No matter how it is referred to, IV turns video into "actionable information," which allows users to receive alerts and make decisions regarding appropriate next steps.
The "intelligence" in IV applications is actually a mathematical analysis of video streams. The data can be used in a multitude of ways, many of which are still under development. The overarching idea is that the surveillance system itself analyzes the video and alerts its operator by triggering an alarm when there is a change to the appropriate level of activity in the field of view. IV is not designed to fully replace human analysis. People will still be needed to assess the entire situation and act accordingly, because human vision is extremely advanced, and is impossible to replicate with mathematical algorithms.
IV can be used in numerous capacities, including object tracking, object counting, license plate recognition, face recognition and object identification. For example, the Boston Police Department has network cameras monitoring the entryway door to their own building. The camera follows each individual as they enter until it gets enough data points for facial recognition. The system then automatically compares this image against an existing database of outstanding arrest warrants. In this way, if someone with an outstanding warrant enters the building for any reason - such as to bail out a friend - the officers know within minutes whether they should detain the person longer.
Offering this sort of intelligence in the video system creates major advantages, the most central of which is the ability to reduce the workload on staff. The IV system is never idle. It is constantly on guard, waiting for an impulse to send an alarm or start recording. There are a number of different ways to set up an IV surveillance system and important factors, like image quality that should be taken in to account.
Surveillance System Architecture with IV
IV can be incorporated into an existing surveillance system, or built into the architecture of a new system. There are two different types of network security architectures that utilize network video. Those two methods are 1) centralized intelligence, in which all intelligence features and algorithms occur in one location, and 2) distributed intelligence, in which the IV functions occur at dispersed points throughout the installation.
Centralized intelligence - This is most common in a system utilizing DVRs to convert and store video from analog cameras. In this type of system, all IV algorithms are housed at the DVR level along with digitization of the video and video management functionality. In this set up, all computing power is centralized in the DVR, which means the number of cameras that can be analyzed is limited, making the system less scalable.
Distributed intelligence - Distributed intelligence can be used in a network video system using analog or network cameras. If analog cameras are already installed, video servers can be added to the system and used to digitize analog video and run IV algorithms closer to the camera level. The processed information is then funneled through a network switch to storage devices and monitoring stations.
In a network video system the edge devices -- the video servers or the network cameras themselves -- have built in computing power to run the IV algorithms, pushing the intelligence all the way to the periphery of the surveillance system. This makes the system scale from one to thousands of cameras without over burdening the centralized recording device, like in the DVR scenario. It also decreases the amount of video sent over the network because the cameras themselves "decide" when recording is necessary. This in turn reduces the overall strain on the IT infrastructure by freeing up bandwidth for other applications.
IV and Image Quality
Along with the mathematical algorithms, image quality is of extreme importance for the accuracy of the IV system. Without clear images, the best IV algorithms will not be able operate accurately. Network cameras bring an end to the interlaced scan problems of analog systems utilizing DVR technology. Interlaced images are created from two sets of lines that update alternately. This delay causes a blurring of the overall image. Network cameras utilize a newer technology to create images called progressive scan. Progressive scan captures the entire image at once, so even with a high degree of object motion, the image is clear.
Megapixel and Immersive Imaging
Analog video systems are tied to television specifications, meaning the maximum resolution is 0.4 megapixels when digitized. Standard digital still cameras available at retail stores are now 5 megapixels and 1.3 megapixel cameras are built into cell phones. Network video cameras can also utilize megapixel technology, which has some obvious benefits, beyond just getting a clearer image. Details from megapixel cameras are more easily recognized in the image. More details means additional data points for IV algorithms, which in-turn improves accuracy of the analytics.
Another way to utilize megapixel technology is for what's being called "immersive imaging". By using a wide-angle lens attached to a megapixel camera, the camera can span a much wider field of view (some camera lenses designs even cover a full 360 degrees) than normal cameras. Immersive imaging facilitates digital pan/tilt/zoom (PTZ). The result is the ability to pan, tilt and zoom in on a field of view, even though the camera stays put. Because there are no moving parts, users don't experience the mechanical wear and tear that exists in analog PTZ cameras which must physically move There's also a potential gain in speed, since an analog/mechanical PTZ can be no faster than its drive motor.
IV, megapixel and immersive imaging offer a number of benefits to an existing or new surveillance system. IV can lower the total cost of a surveillance system by generating fewer false alarms, and by reducing the amount of people required to operate the system. The surveillance system will alert personnel as appropriate when an unusual event occurs. Megapixel imaging allows for even higher resolutions, which in turn allow IV algorithms to act even more exactly.
To be most effective it is critical to work with vendors that employ open standards for the use of IV. This allows the user to choose the best IV algorithms and applications for their needs without having to worry about interoperability challenges.
IV, megapixel and immersive imaging remain hot because they will greatly improve system performance and will continue to evolve creating even greater user advantages in the coming years. Network video is a best of breed system, utilizing open computing platforms and storage systems, which will result in new hot technologies on the horizon faster than usual.
About the author: As the general manager for Axis Communications, Fredrik Nilsson oversees the company's operations in North America. In this role, he manages all aspects of the business, including sales, marketing, business expansion and finance. He can be reached via email at Fredrik.Nilsson@axis.com.