It's not cricket: technology usurps the umpire

Cricket is grabbing front-page headlines for all the wrong reasons. Two of Pakistan's players have overstepped the mark (literally, by misplacing their feet and bowling 'no-balls'), and face allegations of pocketing cash for deliberately doing so at pre-arranged moments in the fourth Test. Bookmakers could have been ripped off by their actions if corrupt punters had placed bets on when the no-balls would occur.

This revelation follows widely held suspicions that there has been corruption in the game for several years, and many cricket lovers fear that this latest betting scandal may represent the tip of the iceberg. What they and the cricket authorities desperately want is media coverage that focuses on far more positive aspects of the game. In their eyes, it is a great shame that the recent England-Pakistan series will not be remembered for marvellous swing and spin bowling, or the fantastic, record-breaking partnership between Stuart Broad and Jonathan Trott.

The betting scandal has also overshadowed another significant event: the successful introduction of a new umpire review system into English cricket. This allows players to appeal against an umpire's decision and have the ruling referred to another umpire sitting in a technology-packed room in the stands. Original decisions may be overturned after the third umpire has scrutinised evidence provided by ball-tracking systems featuring a predictive element and infrared cameras capable of detecting faint contact between the ball, bat and pads.

The International Cricket Council (ICC) introduced the umpire review system just after the end of last year's English season, and it soon played a role in several international games played in the southern hemisphere. However, this was certainly not the first time that technology had played a part in the game of cricket. Since the early 1990s on-field umpires have had the option to refer decisions to a third umpire, who scrutinises several slow-motion replays, and decides whether a given decision is correct.

Fans assume this approach is infallible, but that's not the case. TV cameras generate pictures by raster scanning, starting with pixels in the top left of the screen and finishing in the bottom right. So a frame can show that a batsman has just crossed the line when the bails were whipped off, although in reality the bat was actually a few inches short of safety. But this is not a major concern in the longer term, because broadcasters will soon start to upgrade to cameras capable of far higher frame rates.

Hawkeye

From the armchair fan's perspective, the most important technological breakthrough came in 2001, when a ball tracking system called Hawkeye made its debut on TV screens. This system, which forms part of the third umpire's toolkit, records the path of the ball from the bowler to the batsman. It also features a predictive element that can assist the notoriously fuzzy leg-before-wicket decisions. Hawkeye reveals where the ball landed on the pitch, where it struck the batsman, and the path that it would have taken if the batsman's leg had not been in the way.

Hawkeye is the brainchild of the vision-processing group at Roke Manor Research, a Siemens-owned research and development firm based in Romsey. This group pitched the idea for a ball-tracking technology to Sky TV and Channel 4 in the late 1990s, and the latter commissioned them to build a working system. Sky was alarmed when it discovered that the free-to-air broadcaster would be featuring ball-tracking technology in its cricket coverage, so asked the UK defence technology company QinetiQ to build a competitor.

'QinetiQ's system never worked, and Sky went with us,' explains Paul Hawkins, a former Roke employee who now heads up Hawkeye Innovations.

His company's tracking system is based around six calibrated, high-speed cameras that are positioned around the outfield and focused on the 22 yards between both sets of stumps. Each camera picks up the location of the ball in every frame of video, and a three-dimensional record of the ball's trajectory is constructed via mathematical manipulation of the data. These calculations treat the trajectory from the bowler's hand to its contact with the ground and from the ground to the batsmen independently, but join these paths at the bounce point. By adopting this approach the system is not phased by variations in the bounce of the ball, how far it spins off the turf, or whether it hits the seam and deviates.

Predicting where the ball would have gone if it hadn't been obstructed by the batsmen involves a combination of curve-fitting and application of Newton's Laws of Motion. The predictive aspect is more challenging when the batsman has advanced down the track towards the bowler, and the ball has hit his pad just after it has bounced.

'If it's only 10cm after the bounce, you don't have enough information to get a good feel for how the ball is curving,' says Hawkins. 'So you have to use gravity and things like that.'

It is possible to determine the trajectory of the ball from the data provided by just two cameras. But redundancy is beneficial, because it creates a system that can cope with camera failure. 'If you had only two cameras, and the calibration was slightly off on one of them, you wouldn't know which one was wrong,' explains Hawkins. 'When you've got six you're in a much stronger position, because if one of them is off, it's much easier to see which one it is.'

Independent trials have revealed that Hawkeye's values for pitching and interception points are accurate to within 2.6mm. It is tough to determine the magnitude of the uncertainty for the predictive path because there is no definitive approach for revealing where the ball would have gone if it had not been intercepted by the batsman. However, Hawkins and his colleagues say that this particular error is usually less than one centimetre.

Snickometer

Another new toy for the TV cricket fan was Snickometer, a package develop by UK inventor Alan Plaskett that analyses the sounds recorded by an on-pitch microphone close to the batsman. Obviously this contraption could form part of the third umpire's toolkit, but it was not used in the recent England-Pakistan series. Plaskett knew that the best way to distinguish between the sound of the ball rapping the batsman's pad and hitting his bat was to study its frequency. So he wrote computer code that used a mathematical function known as a Fourier transform to convert the audio signals into this form. Graphs related to these processed audio signals are synchronised with a slow-motion replay, allowing the viewer to determine the source of the noise.

Hot Spot

Plaskett has also teamed up with Warren Brennan, owner of Australian technology company BBG Sports to produce another version of Snickometer for that market. Brennan, a big cricket fan with over 20 years of experience in sports technology, is also the driving force behind the introduction of Hot Spot, a high-speed infrared camera that detects the heating caused by the impact of ball on bat or pad. When batsmen thrust bat and pad forward together it is tough for the umpire to decide what the ball has hit, but Hot Spot can reveal the local heating that results and detect it for up to half a second.

To determine the most suitable infrared camera technology for cricket, Brennan organised a trial at Old Trafford, Manchester. This comparison involved several cameras already employed for a wide variety of other applications: border surveillance; maritime surveillance; nighttime gun sighting from a helicopter; and missile tracking.

The best results came from cameras with mercury cadmium telluride detector chips, which operate in the 3-5m range. This narrowed Brennan's selection to a choice between a camera from the US firm FLIR, or one made by the French outfit Cedip Infrared Systems. He selected the latter.

'One of the key advantages that we had over FLIR was that our technology was European and therefore much easier to export around the globe,' says Jon Chicken, a former Cedip employee who transferred to FLIR when it acquired the French firm. Interestingly, this takeover has had no impact on travel restrictions for the camera, because the export license actually depends on the country of origin of the technology, rather than the distributor's headquarters.

Infrared cameras differ from their visible counterparts by detecting radiation emitted by an object, rather than the light reflected off of it. For temperatures associated with cricket ball collisions, infrared radiation is most intense around 10m. However, Chicken says that better results can be obtained with a camera operating in the 3-5m range, known as medium wave infrared. 'At the peak of the [intensity] curve a small change in temperature results in a small change in the response of the cameras, whereas in the medium-wave you are on the steepest part of the slope, and the same small temperature change results in a bigger difference in response of the camera.'

The imaging hardware is not cheap. The basic version lists for £70,000, a hefty price tag that partly stems from the need for specialist optics transparent at 3-5m. However, by far the most expensive component is the detector chip that is housed in its own sealed package. This assembly also features a mechanism for cooling the detector, which is made of mercury cadmium telluride. Silicon, the material used to make detectors in today's digital cameras, is not suitable because it is transparent in this wavelength range.

Mercury cadmium telluride technology is well developed, and it has been used to build infrared detectors for the military since the 1970s. Initially these imagers used a raster approach, combining multiple, rotating mirrors with either a single element or linear detector, but since the late 1990s they have been superseded with a pixel array.

Image acquisition now involves the focusing of infrared radiation onto a chip based on mercury cadmium telluride. Photons that hit this detector liberate electrons from their bound state, which are then swept away by a built-in electric field to a silicon chip that stores this charge. Every time a frame is generated, charges from each pixel are amplified to form an electric signal proportional to the incident light intensity at that location. Gathering all of these signals can form a snapshot image.

In the firing line

Hotspot, Hawkeye and Snickometer are all subjected to far greater scrutiny than ever before because they are often being used to make umpiring decisions rather than just improving coverage for the TV viewer. Of these three technologies, Hawkeye is receiving the most criticism, and Hawkins believes that this is unfair. He claims that every decision that Hawkeye has been called upon to make by the umpires has been correct, and argues that his detractors, who include several famous former players-turned-commentators, speak unfavourably of Hawkeye because they have failed to take the time to get to grips with the technology. 'Everyone who has taken the time to see the system and understand the system believes in it,' he protests.

Hawkins believes that Hot Spot is the weak link in the technology toolkit, arguing that there are times when it fails to pick up the faint contact between bat and ball that has been detected with other technologies. 'But people can forgive Hot Spot, because everyone easily accepts how it works. With Hawkeye, where the lay person doesn't accept how it works, it requires a greater leap of faith.'

Hot Spot's owner, Brennan, admits that there have been problems with the technology, but these are being addressed. 'At the start of the [Australian] summer against the West Indies we only had two cameras operating at the North and South ends of the grounds. This approach relies on batsmen turning the bat to show where the nick happens, but we found that several batsmen weren't doing that. They were pulling their bat back quite purposefully, and putting it between their legs or behind them.'

More cameras were needed to prevent batsmen from hiding the evidence created by these faint contacts. However, BBG Sports could not deploy its other two infrared cameras straight away, because they were covering the New Zealand-Pakistan tour. However, when that series finished they started using all four on one ground.

'From side-on we got some incredible results. We were running the cameras at their full frame rate of 115 frames per second, and we got some very fine nicks with the side-on cameras that we couldn't see with the front-on cameras.' This four-camera approach was also implemented in Pakistan's tour of England.

Although Hawkeye and the other technologies are not perfect, they are clearly giving the game an opportunity to reach the correct decision more of the time. And the technology will continue to improve, thanks to the development of new packages, such as BBG's HotSnick ' a combination of Hot Spot and Snickometer - that should be ready by the end of this year.

However, although the ICC wants to use technology to make close calls, it is not supporting it financially, and the broadcaster tends to foot the entire bill. The upshot is that not all of technologies are used in every series. For example, TV coverage of the 2009/10 tests between South Africa and England did not include Snickometer and Hot Spot - two technologies that could have helped the third umpire to detect the faint contact between bat and ball. The unwillingness of the ICC to stump up a relatively small amount of cash to help the umpires that it employs from getting more decisions right more of the time seems absurd and inexcusable - put simply, it's just not cricket.