Eyeing the future of video surveillance

While some of us fondly remember record players and VCRs, millions of young adults today have only ever known digital technology. For the iEverything Generation, they might be shocked to learn that the first digitally-based, network video camera was invented only 16 years ago.

Looking ahead, will the market be fully converted to IP? Will the internet play a much bigger role in surveillance? Will analytics of the future finally match what we see in TV and movies today? But before looking into the crystal ball, I found myself reminiscing about the first 16 years of network video and what our predictions were back then.

When we launched the world's first network camera in 1996, I'll admit that it had lousy performance. It did one frame per second (fps) in CIF resolution and took 17 seconds to generate a single D1 snapshot. It was practically useless for normal surveillance purposes. Fortunately, it found initial success in remote monitoring and we saw a future opportunity in a CCTV market that was 100 per cent analogue, yet bound to go digital just like everything else in society.

Shift away from analog

An easy way to forecast the future of video surveillance is to examine Moore's law, an electronics trend that states performance will double every 18 months for the same cost. We see obvious proof of Moore's law in consumer electronics, especially with personal computers and smartphones, but it doesn't stop there. Today's network cameras can do 30 fps in HDTV 1080p resolution compared to one fps in 0.1 MPix 15 years ago – a 600 times performance increase. This means that network cameras have actually outpaced Moore's law, and today offer far better benefits than analogue.

These benefits of IP video are clear when it comes to image quality, system scalability and ease-of-installation, specifically when installing systems of more than 25 cameras. This tipping point will drop significantly thanks to hosted video (which I'll discuss later).

The shift from analogue to IP has accelerated through the recession, as both manufacturers and consumers focus their spending on technology of the future. However, much as consumers might be surprised to learn that Sony was still manufacturing the Walkman up until October 2010, we will still see some analogue cameras being sold. But, as today's young adults grow into professional security roles, the arguments for analogue technology will become worse and worse, not only because network cameras are easier to install and provide better quality, but because this digital-only generation will expect the continued progress that IP can offer.

Move towards standards

We've seen the importance of proper standardisation in all industries for large-scale technology adoption. Supporting good standards leads to ease-of-use, which is one of the reasons analog video has remained dominant for so long.

Because all the major IP surveillance players have invested in ONVIF support (including my own company), I think it will be the dominant standard on the API for networked video. I also expect PoE, HDTV and SMPTE standardization to continue to have a major impact in video surveillance.

Image quality and HD resolution

Even though network video provides much better image quality, there is still a lot more to be done. In the last 15 years most of the evolution has been on resolution and frame rate. In the future, Moore's law performance improvements will be used for image processing. I expect that this will enable the average surveillance camera to see more than the human eye – something that can't be said today under hardly any circumstances.

This is good news for our industry as we all benefit from continuous investments in R&D. From a technology point of view the shift has been from tube-based cameras (Anyone remember those?) to CCD sensors and then to CMOS. In 15 years, CMOS will likely dominate but we might see another emerging technology. New CMOS technology will create sensors with huge resolutions that will lead to the first Terapixel camera. When this happens, it will be the optics, not the sensor, which set limitations on image quality.

Additionally, for the past 70 years we have lived with the analogue standards of NTSC and PAL. Today, nearly everyone has HDTVs in their homes. As a surveillance professional, I would expect better image quality at work than at home – not the opposite. HDTV is perfect for surveillance because the SMPTE standard guarantees frame rate, resolution, color fidelity and aspect ratio. While megapixel is a trendy topic, it simply refers to the number of pixels in the image – all those other factors of a moving image are variable. This is why the Best Buys of the world talk megapixel for still photography and HDTV for home entertainment.

I don't expect the HDTV standard to last for 70 years, but 15 years from now it's believable that the majority of cameras will be HDTV compliant. Having said this, we will of course see multi-megapixel (and Terapixel!) cameras play important roles – either to store video in higher, more detailed resolution for forensic review or to crop out individual HD-streams. The beauty of network video is that it does not put boundaries on resolution.

Thermal imaging

Everyone wants to have the lowest possible Lux rating for their cameras – but how about having zero? Today thermal imaging is a specialty market, mostly found in military and government applications. As the prices of thermal network camera components decrease and demand increases, we can expect a lot of new applications to arise. Today there is one thermal camera per 400 regular surveillance cameras in circulation. We expect this ratio to reach 1:50 in a few years as surveillance professionals realise that this technology is affordable, can be easily connected to their existing network system and used for many critical applications.

On-board and cloud-based storage

Just like network camera technology has done since 1996, the storage market – including flash memory and Hard Disks – has also outpaced Moore's law. Soon we will have on-board storage in cameras capable of recording HDTV resolution for weeks at a time. This will be a game changer as the camera becomes the recording device, something that can't be done using analog. The people who will have to adapt to this change are VMS manufacturers and hybrid DVR guys. DVRs as we know them will likely go the way of tube-based cameras and will be extinct 15 years from now.

Hosted video

Today, consumers have all learned to rely on hosted services like Hotmail, Gmail and Facebook. For professionals, many rely on Salesforce.com and cloud-based HR systems. And we all rely on internet banking. If we can trust the cloud with our money today, it's logical that we would do the same for security video tomorrow. The benefits are obvious: no need for a DVR, the option to have local NVR recording and no fixed cameras counts, coupled with the basic benefits of IP video. You can view the video from anywhere on Internet-enabled devices, including your mobile phone. Hosted video is a prediction I see having an immediate impact over the next couple years – especially in the target market of small businesses – and then growing to extremely high user levels by 2025.

Cameras with the installer and user in mind

As the iPhone revolutionised how cell phones are designed, networked video is poised to do the same to the CCTV industry. While the shift is often slower when compared to the consumer electronics industry, further improvements to installation friendliness, flexible mount options and a higher degree of PTZ cameras can be expected. Guards will also need to be trained and equipped with mobile devices that connect to the cameras, which will allow for more efficient usage.

Analytics

This is a dangerous area to predict and one where many have been proven wrong for the last 5-10 years. I'm confident in saying, however, that in 15 years analytics will be mainstream. The difficult part is pinpointing when exactly it will happen.

Clearly, as processing power improves, we will see both analytics on the server and on the edge (in the camera). The winners will be those who partner with the best-in-class analytics providers and open-based systems. Having analytics run on the edge scales better, so I expect that most future analytics will happen there by feeding the VMS with either Metadata or alerts. Crucial to the success of analytics companies is that they have the end-user in mind through the entire development process – as opposed to bundling fancy analytics with every camera and hoping the installer sorts it out. Yet again, Apple is on to something with the AppStore model. Expect the surveillance market to follow.

Industry verticals and trends in society

Today, we see most surveillance cameras in retail. As network cameras continue to improve, integrators and their users will discover many more applications. City surveillance, transportation and healthcare verticals will experience increased growth as surveillance systems enable better operational efficiencies. The residential market is another to keep an eye on. While many have tried selling cameras to homeowners, I expect that this will become a larger vertical thanks to hosted video.

We will see more regulation on how the video is used, but less on where cameras are placed. The cost of deploying cameras systems will fall – mainly in the storage and installation sides, but also likely with respect to software. Given the overall growth in our industry and the new applications that technology makes possible, I think we will see a lot more cameras in new vertical markets where we never expected them. This means a bright future for everyone in the industry!

Martin Gren is the co-founder of Axis Communications AB and the inventor of the world's first Network Camera.