A quarter of the European machine vision industry’s sales are now for non-manufacturing applications, and this figure is set to rise year-on-year. Machine vision is at the forefront of new, highly visible systems and initiatives, such as automated security systems (see the feature on page 17) and road traffic management, as well as ‘hidden’ systems such as address recognition in letter sorting machines used by postal services. We tracked down some key players from the machine vision industry to gauge their opinions on the latest expansions beyond manufacturing and production.
Don Braggins, of UKIVA said: ‘For a long time we’ve been telling our members that the technology we have in the machine vision industry should be used in other areas, such as automated security. We’re now starting to see machine vision players expanding out of the factory into new applications.
‘One major area for machine vision is in transport, particularly traffic monitoring. Systems are now in use that can recognise different vehicle shapes and read the registration plates. This information is very useful when traffic and congestion management is a particular concern. For example, technology for recognising number plates is vital to the administration of the congestion charging in central London. A very similar system is being used on major roadways in Britain to provide highway managers with information about traffic flow, such as average journey distance and time.’
Nick Hewitson, managing director of Smart CCTV, highlighted how machine vision is changing the world of surveillance: ‘The major problem with standard CCTV is that it’s not really effective at preventing crime. The current method is to record the video and hope that the image quality is good enough to allow the identification of the culprit – but even if this is successful, the crime has already been committed.
‘The introduction of behavioural recognition into the security world offers a major change by highlighting the possibility that something is about to happen, giving the police and security staff more time to react and, therefore, catch the perpetrator in the act. In turn, this changes the way criminals perceive the risk of being caught, thereby reducing the incentive to commit an offence.
‘Behavioural recognition software can monitor how people move within the camera’s field of view and looks for unusual behaviour patterns. The software then delivers this video source to the desk in the CCTV control room, thereby allowing the members of staff to deal with actual and potential incidents, rather than trying to watch many monitors all at the same time.’
In many sports, balls move with incredible speed: tennis and cricket players can propel balls at opponents at speeds upwards of 90 miles per hour. With balls flying around at such a pace it can be very difficult for umpires to make potentially match-changing decisions with absolute certainty. Fortunately, a technology heavily reliant upon machine vision is changing the way that borderline decisions are made.
Hawkeye uses a series of six cameras stationed around the field or court to follow the flight of the ball. By analysing the information from all the cameras the system not only traces the speed of the ball as it changes through its flight and how the ball rebounds from the bat or racquet, but also measures the spin of the ball and the way the ball bounces from the ground. Hawkeye can even trace where the ball would have gone had it not been intercepted.
All the information gathered by the cameras and processed by the software is replayed in a ‘virtual world’ where television viewers are treated to various interpretations of the game as it unfolds in front of them. The system also generates a wealth of statistics, along with a detailed replay system that can be used by players and their coaches to improve the standards of their game.
Automated vision systems are also finding uses in environmental applications, particularly detecting potential fires in waste storage areas. The modern consumer society we live in generates an ever-increasing mountain of rubbish. Many advances have been made in reusing and recycling household waste, but inevitably there is still a large amount for which there is no alternative other than incineration or burial.
Prior to waste being incinerated, it is often stored in so-called waste bunkers. In these bunkers pressure can build up, increasing the temperature of the waste. Spon-taneous chemical reactions can occur, and methane can accumulate as part of the decomposition process. All of these factors can lead to an increased risk of fire. Such a fire is both hazardous for workers at the waste management site, and a potentially serious environmental disaster; dangerous chemicals can be released into the atmosphere through uncontrolled burning, and water used to fight the fire can become contaminated with harmful substances that are soluble.
German company m.u.t designs and installs systems that automatically detect potential fires before they have a chance to become established. It uses thermal cameras from Flir to look for telltale hot spots in the waste. Because the cameras see in the infrared, they are unaffected by smoke and dust in the atmosphere in the outbreak of a fire. Using the video images, the ARTUS software determines the current temperature values of the surface monitored and records the measured values. Hot spots hidden under the surface are also detected as they cause an increase in surface temperature long before an open fire starts.
If an abnormally hot area of waste is discovered, the system alerts the monitoring operators to the potential risk by setting off an alarm and displaying an image of the suspect area. In case of an alarm, the user can switch from automatic to manual mode to make a detailed analysis of one single zone. The operator can then take appropriate action to prevent a fire developing.
These are only some of the novel ways in which smart machine vision systems are now being used away from the traditional areas of automated production. As the technology becomes even more advanced, the range of new applications for machine vision will grow exponentially.
