Thanks for visiting Imaging and Machine Vision Europe.

You're trying to access an editorial feature that is only available to logged in, registered users of Imaging and Machine Vision Europe. Registering is completely free, so why not sign up with us?

By registering, as well as being able to browse all content on the site without further interruption, you'll also have the option to receive our magazine (multiple times a year) and our email newsletters.

All-seeing eye

Share this on social media:

Topic tags: 

Stephen Mounsey takes a look at the cameras that are taking a look at all of us

Recent tabloid headlines have claimed that the UK has as many as five million CCTV cameras monitoring a population of around 65m – far more than the whole of mainland China has between its one billion inhabitants. A large majority of those five million cameras are analogue systems, built on out-dated technology and delivering poor image resolution by modern standards. As customers in the security and surveillance industry gradually transition to digital equipment, some manufacturers from the machine vision sector are finding that their expertise is in demand.

While the UK has the most extensive surveillance structure in the world, UK customers fall behind in terms of installing up-to-date solutions. According to estimates by one market research firm, some 70 per cent of new surveillance cameras sold to UK customers and integrators are analogue, whereas analogue systems account for just 30 per cent of worldwide sales. Replacing older technology with new will represent a significant market for some time to come – a market which some machine vision suppliers are now entering.

But why should customers switch to digital systems? What advantages do they offer? Rick Ramsay is a product manager at Avigilon, a Vancouver-based designer and manufacturer of HD surveillance systems. According to Ramsay, the chief attraction of digital, networked cameras is that they lead to a high resolution system: ‘Our entire approach is all about capturing, storing, and managing better quality video, and getting over the hurdle of the low quality of CCTV video we’re all used to seeing on the news.’ The company’s systems are aimed at customers who are ‘actively using’ their surveillance systems, he explains. ‘The goal of bringing HD into surveillance is to move the technology from being a story-teller to being a provider of video evidence; rather than just being able to say “a man in a red sweater robbed you”, we can now see his face clearly. The video then becomes the primary investigative tool, rather than just additional evidence.’ In Avigilon’s case, the ability to provide usable evidence is enhanced by use of an encrypted hash value, stored with every recording in order for later validation of the data’s integrity.

Similarities and differences…

In digital surveillance systems, Ethernet cables can power, control, and provide bandwidth to networks consisting of hundreds of so-called IP (internet protocol) cameras. As a machine vision company that has recently made a move into the surveillance market, Basler Vision Technologies is well placed to contrast the different demands made by the two application areas. Hardy Mehl, director of Basler’s IP camera business unit, explains that security applications are just as demanding as machine vision, but in different ways: ‘In both industries, customers are ultimately interested in high image quality. The big difference between the technologies is that in machine vision you have an objective image quality, which means that the camera should give you real world information, but in the security world it’s a little different; customers here want subjective information. If you’re looking at an outdoor scene for example, the end customer (a security guard for example) will like the camera if the colours are more impressive in the image than in real life, whereas this wouldn’t be acceptable to a machine vision customer.’

Additional differences exist in the way in which data is transmitted: ‘Data coming from machine vision cameras is almost always transmitted in a raw data format for computer processing, whereas all data leaving the camera in a security application is compressed data – the camera itself runs a compression algorithm.’ In machine vision, it is the transmission of raw data that necessitates the specialist high bandwidth interfaces such as GigE Vision and Camera Link. Additionally, he says, security systems may record 24/7, meaning that storage of uncompressed video would quickly become expensive. ‘Compression removes the interface bottleneck seen in the industrial world,’ he adds.

When specifying just how well a digital security system needs to perform, Mehl says, analogue systems serve as a starting point: ‘The customer wants live images at roughly 25 frames per second, mainly because that’s what he’s used to getting from analogue cameras. As well as this, he wants to increase from PAL or NTSC to 1,080p, which is six or seven times higher resolution,’ he explains. These HD cameras need to be able to compress a lot of data in real time on the on-board chip, he says, adding that Basler develops security application-optimised compression chips, capable of maintaining high frame rates at high resolution.

In a parallel universe

Readers of Imaging and Machine Vision Europe might not have heard of Swedish networked camera specialist Axis, but the company is the biggest IP camera supplier in the surveillance market, with a €325m annual revenue. ‘In the industrial world, people don’t know this company,’ says Basler’s Mehl. That, he says, is a good demonstration that despite using similar cameras, the two markets are very different from each other.

Martin Gren, co-founder of Axis, recollects that the company started in the business of producing equipment to network printers, before releasing one of the world’s first networked cameras 15 years ago: ‘That first camera was capable of just three frames per minute, and was therefore hardly suited for surveillance and not at all suited to machine vision applications,’ he says. The company, which designs its own chipsets, went on to push the performance of its networked cameras to 30 frames per second, which Gren notes is the minimum requirement for CCTV.

Gren highlights unpredictable lighting conditions as a particular difficulty: ‘In machine vision you usually have a well-controlled light environment, but in the surveillance industry you never know what the lighting will be like, and this is a significant difference, making image processing very challenging. Shadows, reflections, and an unknown direction of lighting are all to be expected in surveillance applications, particularly outdoors.’

Although the movement from analogue to digital surveillance could encourage more machine vision companies to follow Basler’s lead, Gren is not worried by the prospect of increased competition. ‘We have learned that there is a different kind of image processing that we want to do in the surveillance industry compared to a what would be done on a handheld digital camera, and I would assume there are different things you want to do when it comes to machine vision compared to a surveillance camera,’ he says. ‘If you take a consumer handheld camera, for example, it’s supposed to take a nice image of people who want to be photographed, and if it’s dark you use a flash. On a surveillance camera, however, if you want a light it has to be a lamp, and that’s going to take a lot of power. Also, the people it’s trying to photograph are moving, and may even be trying to hide. This all adds to the requirements placed upon the image processing.’

Gren adds that Axis keeps ahead of any competition through active product development, having recently launched one of the first IP-networked thermal imaging cameras. Although these cameras are significantly more expensive than their regular counterparts (starting at around $4,500), a single camera with appropriate processing software is able to cover a very large field of view. ‘These are finding applications in transportation, in railway yards, high security infrastructure, and we’ve also sold quite a few to the education sector; one camera can cover a whole school yard, replacing a great many PIR detectors in detecting intruders.’

Educating the masses

‘Although the market is rapidly changing to IP camera technologies,’ says Basler’s Mehl, ‘and although growth is at 30 to 45 per cent per annum, there are still some obstacles reducing the growth rate. One big issue is that if you already have an installed analogue solution, and you want to renew the cameras or extend the security infrastructure, you end up with a problem that you have to re-wire the existing infrastructure completely.’ This is typically a very expensive process, he says, and where security systems are refurbished or extended (rather than being built from scratch), a price-performance comparison will often favour retaining analogue technology. Digital signals can, he adds, be transmitted over the coaxial cables of an analogue network, but this is also expensive.

The difficulty of setting up a digital system may also influence its uptake: Avigilon’s Ramsey comments that customers in the security space are used to setting up analogue devices, which was a relatively straightforward process. For this reason, he says, Avigilon puts a focus on designing its systems to be as simple as possible to set up, with components self detecting and configuring automatically.

Axis’ Gren believes that the main obstacle to increased market share is a lack of understanding among security and surveillance customers as to what a networked video solution can offer them: ‘What we’re working on at the moment is educating the market – to tell all of the security professionals who keep on buying analogue, because they know no better about the quality and benefits of networked video.’ Perhaps as a result of such education, IMS market research has seen a large increase in IP surveillance camera sales in Axis’ native Scandinavia, where 50 per cent of cameras sold are now networked devices. In certain other markets, such as parts of the Middle East, IMS reports similar figures of greater than 50 per cent of new cameras being IP-networked.

While the surveillance applications may borrow from the machine vision space, it is the industry’s more distant cousins in consumer electronics that are likely to drive adoption of networked digital surveillance systems in the future: ‘I think the most interesting thing is that a security professional may have HDTV at home, and may be able to watch a football game in full-HD with his family, but then when he goes to his workplace he often has to settle for the older television technology. This is kind of strange, considering he needs to be able to see details for his work,’ says Gren.

One area in which digital technologies will always win over analogue is that of storage, and lower storage costs. In analogue systems, says Gren, four, eight, or 16 cameras make up an isolated island, which will require its own recording system. In contrast, Axis has carried out a project for a large US retailer in which 100,000 video channels were integrated.

Finally, Basler’s Mehl believes that machine vision players could be more aware of the convergence between their industry and that of security and surveillance: Many manufacturers and sales partners in the industrial camera market, he says, think that they know the security market, but traffic control is often the only area of which they are aware, and security applications slip by unnoticed. Perhaps greater awareness could lead to more crossovers in the future.

Other tags: 
Company: 

Related features and analysis & opinion

22 February 2019

Ron Low, Framos head of sales Americas and APAC, reports from Framos Tech Days at Photonics West in San Francisco where Sony Japan representatives presented image sensor roadmap updates