Leave nothing to chance

The Government’s proposed Gambling Bill has caused no end of controversy since its constituent parts were first published in draft form in November 2003 (following on from the independent Gambling Review Body’s report of July 2001 and the ensuing White Paper, which emerged in March 2002).

Initially, the Bill – brought to the Commons by culture secretary Tessa Jowell – proposed the introduction of a Gambling Commission with wide-ranging powers to halt the infiltration of crime within the gaming industry. There were to be strict suitability tests for all commercial gambling operators, stringent controls on access by children to places where gambling occurs, an industry-funded Gambling Trust to support gambling addiction treatment programmes (with a starting budget somewhere in the region of £3 million) and stern licensing conditions.

Hardly any of us would argue that this Parliamentary Bill isn’t desperately needed. The nation’s gambling laws date back to the 1960s but, in the intervening period, the nature of the industry has changed quite considerably. Gambling is now a diverse, vibrant and innovative industry. It’s also a hugely popular leisure activity enjoyed in various forms by millions of people the length and breadth of the country. Inevitably, the law must reflect that state of affairs.

However, following scrutiny by a Joint Committee of both Houses of Parliament – which numbered Richard Caborn MP, the Minister for Sport and Tourism, among its ranks – Britain’s gambling industry moguls have roundly dismissed the Government’s proposed deregulation as a thinly-disguised attempt to attract foreign investment at the expense of the country’s existing gaming companies. Both John Kelly (boss of private operator Gala) and Rank Casinos’ chief David Boden are railing against Labour’s proposals which, they suggest, are seriously flawed and merely aimed at helping to attract US giants including MGM Mirage and Caesars Entertainment to build mega casinos costing anything up to £200 million each.

Would such developments in targeted parts of the country lead to regeneration? That’s the official line Labour has adopted to date. Cynics might say that regeneration is a fine building block on which the party in power might construct its General Election manifesto. Aiding the revival of run-down parts of Britain is definitely a laudable objective for Prime Minister Tony Blair, who only recently told the House of Commons that between 20 and 40 casinos would be added to the existing 120 outlets across the country as a result of the Bill.

Government forced to back down

Of late, though, that mounting pressure from the British gambling industry – not to mention alleged reservations concerning ‘super casinos’ voiced by none other than Home Secretary David Blunkett, Deputy Prime Minister John Prescott and Chancellor Gordon Brown – has inevitably forced Blair and Jowell to backtrack somewhat.

Last November, Richard Caborn stated that only eight super casinos would be permitted during the first wave of gaming deregulation. Caborn suggested to Parliament that ministers had taken note of concerns that the licensing controls proposed in the Bill, working alongside the planning system, would not on their own be strong enough to guard against the proliferation of regional casinos in unsuitable areas. Politicians listening to the rank and file? Whatever next?

The Prime Minister is adamant that more casinos in the UK will not increase organised crime, and has dismissed as “complete nonsense” the notion put forward by the Conservative’s culture spokesperson John Whittingdale that the Government had already engaged casino bosses on the topic of tax rates. No commitment has been made to a reduction in taxes on casino operations, but it’s likely that tax rates in the 40%-50% band would likely deter any investment from the US.

The current Government climb down has also resulted in restrictions being imposed on advertising, with a ban on posters promoting online casinos in railway stations and on the London Underground network.

In another possible compromise, Viscountess Cobham – current chairman of the British Casino Association – has proposed a three-year moratorium on slot machines with unlimited prizes (the so-called Category A machines) to enable more research into the possible damage they cause. In an open letter to MPs, the Viscountess has also suggested that membership should be required for entry to all casinos, and that a separate planning class be introduced for regional developments.

Given that the UK’s gaming laws are obviously past their sell-by date, could it be the case that systems provision and installation management might need an overhaul as well? Certainly, the casino environment is a harsh and demanding home for surveillance systems, with dimly lit spaces in direct contrast to the vibrant ‘blocks’ of colour atop the roulette tables. There’s much activity in confined spaces as croupiers deal the chips and spin the wheels while gamblers throw down their gauntlet (in the shape of pound notes or dollars) and drink their nerves away.

Traditionally, security managers in the gaming sector have shied away from ‘going digital’, but maybe – just maybe – that situation’s about to change. Largely thanks to the American influence. In the United States, gambling is controlled on a State-by-State basis, usually falling under the remit of the State Gaming Commission. Given the history of organised crime that has pervaded the industry Stateside, it’s no surprise to learn that casinos across the Pond are highly regulated. The regulators take an active interest in the probity of all suppliers (and their suppliers, too), which includes full background checks on all company directors. That’s as it should be.

In addition to these checking procedures, the regulators are interested in the technical performance of the system itself, and will often specify requirements for digital recording, right down to the frame rate (which, in most cases, is effectively a full frame rate of 30 frames per second – or fps – in NTSC, 25 fps in PAL format). While the National Indian Gaming Commission (NIGC) stipulates a minimum of 20 fps and images of “sufficient coverage and clarity” to identify people and determine the configuration of wagers, card values and game outcomes, it should be noted that 20 fps is an impractical frame rate for digital recording.

For suppliers, the regulatory environment in American casinos means that it’s just as important to work with the regulators from Day One of a project as it is to work with the client. This can include the loaning of security equipment to the end user and the Gaming Commission’s representatives by way of assuring them that the equipment to be installed is indeed fit for purpose.

In America, casinos are more akin to leisure palaces. Huge gaming floors resembling a strange hybrid of a hotel foyer and a shopping mall are a far cry from the image of casinos popularised in James Bond movies. They’re like small cities, offering a range of activities from table-based games through to slot machines that entice Joe Public to relinquish his cash.

Around the gaming tables

Within these vast buildings, which often comprise tens of thousands of square metres of floor space, there are a number of technical challenges to be overcome which – necessarily – are really a function of the area to be monitored. Around the gaming tables, regulators generally require CCTV to be recorded at full frame rate such that it will capture anyone attempting to cheat by sleight of hand. Mostly, this consists of people palming chips, dropping chips on the table after bets have been closed and adjusting chip positions. Basically, it’s all concerned with who did what and when.

Slot machines occupy a far bigger area than roulette tables in the US. With thousands of these machines whirring away, and the occasional slew of jackpot coins bursting all over the floor, the most important element for the security manager is to monitor the area in question for low level crime, pick-pocketing and general public order offences. The slot machines themselves will be tamper-proof and alarmed. In these areas, the camera frame rate is negotiable with the individual Gaming Commission based on a threat assessment.

As you’d expect, all of the financial transactions in casinos are recorded. Threats typically include disputes over how many chips have been given out, how much money has actually handed over as part of a transaction and threatening behaviour towards staff. Chip snatching is not an uncommon occurrence, either. Plus there’s the ever-present danger of pick-pocketing and robbery. The important issue here is the number and positioning of cameras to ensure that all activity is monitored – and that, when an incident does occur, a clear picture of events may then be constructed.

The foyer of any casino is a key monitoring area. Casino managers like to be able to see who’s coming and going such that they can identify valued customers and known troublemakers in equal measure as soon as is practicably possible. In American casinos, the management has to protect the operation against deliberate ‘slip and fall’ injuries in which the ‘victim’ sets out to injure themselves in the hope of receiving financial compensation from the establishment.

In this instance, CCTV should be able to provide recognisable images of individuals entering the building. When transmitting images over IP, it’s vital to ensure those images aren’t overly compressed, and that they possess sufficient resolution to render a clear, recognisable picture.

Security behind the scenes

Behind the scenes there are a number of security issues which may be addressed by CCTV. The obvious area is cash rooms, where those all-important receipts are tallied. In a casino, chips are every bit as good as cash, and the vaults where these are stored have to be treated with as much care as the cash rooms themselves. In some cash rooms, the regulatory authority may specify that audio be recorded in addition to video.

If you’re thinking about designing an IP-based digital surveillance system for a casino environment, the first point to be considered is whether your cameras are PTZ or fixed in one position. Wherever an IP-based system deploys motion codecs such as MPEG, PTZ cameras can cause a problem because of the rapid change in scene from one frame to another. The result is a high bit rate, MPEG artefacts, loss of resolution and the loss of video frames… or maybe a combination of all of these outcomes.

IP cameras and their analogue cousins don’t behave in the same way under similar conditions. If you take the case of a person being tracked across the casino floor by an analogue PTZ camera, you’d expect the background to be blurred (but that the person concerned would remain mostly in focus because they’re being tracked against a stationery background).

With an IP camera, the rapid increase in new scene information can – if the bit rate is set too low – overload the MPEG encoder, causing not only the background to break up but also the image of the person being tracked.

The quality of rendered images can mean different things to different security managers. One can define it as the faithful reproduction of fine detail in an image. Those fine details might be lost due to a number of factors, not least image compression, noise filters and display equipment. Image sharpness may be described in terms of pixel dimensions, and image resolution measured in TV lines. A low number for either will mean poorer quality images.

Pixel dimensions in video is sometimes described in terms of CIF, 2CIF and 4CIF, with 4CIF being the highest resolution (640 x 480 pixels in NTSC format). Depending on the bit rate of the video stream, 2CIF exceeds VHS quality and 4CIF exceeds S-VHS quality images.

One of the tricks that manufacturers use when demonstrating digital recorders is to quote a high display resolution – in other words, an image with high pixel dimensions – while sacrificing image resolution or sharpness which is measured in TV lines. By doing so, they can economise on the bit rate and disk storage space, making their digital image encoder appear more efficient than it actually is. A simple test, using a video signal generator, will reveal the true sharpness of the image as measured in TV lines.

The ramifications of poor quality images and the inability to resolve fine details can be severe in a casino environment. You want to be able to identify a face, read the suit on a tarot card, or determine the configuration of a wager at the roulette table.

One of the great advantages of IP versus analogue – and one which is often overlooked in the debate – is the fact that IP network cameras free the security manager from the constraints of TV resolution. The Axis 206 camera, for instance, is a low-cost 1.2 megapixel camera rendering end users a pixel dimension of 1280 x 1024. That’s about four times the number of pixels of a TV camera, or twice the resolution.

Industry predictions suggest that this will become increasingly important in the future. Slowly at first, maybe, but already higher resolution IP cameras are finding niche applications (as has been shown in SMT), and the ability to drop one of these onto the network simply by swapping it out with a lower resolution model must surely be recognised as a distinct advantage.

Facial recognition: the facts

Facial recognition has been much in the news of late, as experts try to design add-on systems for CCTV systems that can recognise peoples’ faces. The truth is that this is hard enough for people to do, let alone a desktop PC.

One of the biggest hurdles to be overcome in terms of recognising faces is intensity resolution, or image contrast. If truth be told, the human face is a fairly bland canvas. Our eyes rely upon subtle clues of highlight, colour and shading to determine the topology of the face and recognise one from another. Recognition is best achieved under ideal conditions of lighting and proximity to the subject and, despite claims to the contrary, there are no real ways to cheat this truism.

There are hundreds of variables which can influence intensity resolution, from the camera itself to the lens, ambient lighting, shadow, hats, atmospheric conditions and image compression. Something as simple as looking through a glass partition can significantly reduce the contrast values on a face, mainly because of reflections and glare from the glass. Again, codecs can make this worse if the cameras’ compression settings are too high.

If you were to use the example of a person sitting in a car, what happens if there’s some glare coming off the window? At low compression, the codec makes a distinction between the subtle differences in colour and can render a fairly clear image – you can certainly make out the person’s face. However, at high compression rates that extra information about the intensity of the colour (as well as differences in the hue itself) is thrown away, leaving you with nothing but a blob which may or may not be a person.

Nearly every element of a CCTV system – cameras, lighting and compression – conspires to degrade the quality of digital video surveillance. One of the key problems of digital video is there are no simple quality numbers for measuring the output. Describing a system as 30 fps NTSC (or 25 fps PAL) isn’t enough to explain image quality, because in order to achieve that frame rate you may sacrifice quite a bit in the resolution – in respect of spatial, colour and intensity variables.

The trick that manufacturers sometimes use here, when demonstrating their equipment, is to test one parameter at a time (such as quality, frame rate or storage capacity). For instance, greater recording time may be achieved by reducing the frame rate and the image quality. It’s open season on the part of manufacturers to place their system in the best possible light (literally!). By carefully selecting their test scenarios and fine-tuning their systems, they can achieve results that would otherwise be impossible in the real world. The impossible must be possible in the casino environment!

Examining the video evidence

One of the impacts of digital video is the creation of a new class of evidential issues – the artefact. Artefacts are quite simply visual information that has been added to the video sequence and which wasn’t in the original image as seen through the camera lens.

Between the lens and the recording device or monitor, something has been added that means it isn’t a true and faithful reproduction of the scene. In analogue video, this takes the form of noise, speckles, hum bars and so on. In digital video, deteriorating images tend to turn “blocky”, which is caused by loss of colour resolution and contrast. Or they can pixelate (essentially they lose spatial resolution) or show chunks of repeated image from a previous frame of video.

Digital artefacts are caused by finely detailed or fast-moving images coming into the video encoder overloading its capacity to turn images into a bit stream. The encoder copes by discarding information across an ever wider range of pixels. This is a problem with both MJPEG and motion codecs such as MPEG, but the problem is compounded by MPEG wherever the manufacturer is using conditional refresh.

In fact, the worst artefacts happen with conditional refresh. Conditional refresh simply means that if part of a frame doesn’t change very much, that information is not sent by the encoder. This results in savings of both disk space and network capacity.

With conditional refresh, the encoder tells the decoder to assume that part of the frame, or even the entire picture, has not changed since the previous picture. At least not very much. And therein lies the rub, because it is up to the encoder to decide how much is “very much”. If the colour and intensity resolution is set to low, it simply won’t notice that there’s a person creeping along in the shadows at night-time… and because it doesn’t see it, it won’t show it to you, either.

There are many proprietary codecs on the market which are not ISO-standard MPEG4 codecs but carry the MPEG4 label. There are also numerous systems where such codecs were very poorly adjusted, with laughable results. In one system, a demonstration video shows a girl raising her arm above her head. She then walks off but the arm is left hanging in the air behind her! In another example, a car drives past a lamppost and the brake lights come on momentarily. As the car continues past, the lights are seen to jump from the car to the lamppost and stick there for a moment…

In a casino environment, where the veracity of the video evidence is essential to fighting fraud and maintaining credibility, you simply cannot allow something like this to occur. Particularly in respect of table-based games such as roulette, where you have to be sure that what you’re seeing is really happening. Specifying conditional refresh when using a motion codec is generally not advisable.

When the salesperson demonstrates his or her system, make it perfectly clear that you want to be shown the same equipment you’re intending to buy. Examine the specifications to ensure the system matches what you have been shown, and that it’s configured in the correct way.

Always remember that with digital recording, system performance is dependent on environmental factors, so always try to be present at a demonstration in similar conditions to where the cameras are actually going to be located in reality.

Something more than hardware

In the United States, if the CCTV system fails in a casino then that casino cannot trade. If there’s a hardware failure, it’s a crucial business issue because the regulators take security very seriously indeed.

Many DVR systems rely on a central database to index all the images. When you want to retrieve an image, your request is routed to the central index which tells you where, on the myriad of hard drives, your desired image is located.

You can imagine what happens if that database server fails. You’re effectively blind, and what’s worse, you may not even be able to recover that lost information. A truly distributed system won’t allow you to go blind. The information is shared between a number of servers. If a server fails, you lose a small number of cameras for the time it takes you to reconfigure the system.

Software is the key to a robust digital network video recorder. Good software can cope with errors. At the end of the day, the security manager must ask how much investment has been made in the software, because a good digital recording system should be something more than a mere collection of PC hardware.