Imaging and Machine Vision Europe: December 2007/January 2008
The failed terrorist attacks on Glasgow International Airport, Scotland on 30 June 2007, in which a burning car drove into the glass doors of the main terminal building, were just the latest signal that security and transport control must now go hand in hand.
It should be no surprise, then, that the technologies used to ensure security and to monitor and control traffic have converged in recent years, with intelligent CCTV cameras now combining licence plate recognition with complex video analytics to help security staff track potential troublemakers before a crime has even occurred.
This union is evident in a new system released by Smart CCTV, which scans the underside of vehicles to search for explosive devices. The system can also function in conjunction with automatic number plate recognition (ANPR) systems to provide more secure identification of vehicles entering and leaving restricted zones.
Typical applications would include high-security areas such as airports, embassies and nuclear power stations. In these situations registered vehicles are allowed to enter certain parts of an airport where other vehicles are prohibited. To get past an ANPR system functioning by itself, suicide bombers could steal a number plate from a registered vehicle to place it on their own vehicle. However, the underside of a vehicle is unique to its manufacturer and model, and Smart CCTV’s system would deny access if this information does not agree with the data from the licence plate.
As with many of these applications, the system must function for 24 hours a day, often in adverse weather conditions, so the equipment must be very rugged. Positioning the camera to obtain a good view of the underside of vehicles was also a challenge for the company: it needs to be placed at a distance from the vehicle to achieve a large field of view, but it would be unwise to bury the equipment too deep underground. To overcome these problems, the system effectively views the vehicle through a periscope, allowing the camera to be placed in parallel with the road. Once the image is captured by an area scan camera, it is then matched to a database of known vehicles.
In addition to this specialised system, Smart CCTV also provides more general security systems, which make use of video analytics technology to track vehicles and detect suspicious behaviour in addition to reading number plates.
In common with many security and traffic systems now, instantaneous information is provided. ‘They give real-time information to which people can react,’ says Nick Hewitson, managing director of Smart CCTV. ‘The software extracts people and vehicles from the footage to attract the attention of the security guard.’ These systems may draw the guard’s attention to a car that is moving slowly compared to the other traffic, or a vehicle that has stopped on double yellow lines.
One of Smart CCTV’s systems – the ANPR Dolly – is portable, providing a flexibility that makes it suitable for diverse applications. The system would be equally suitable for councils monitoring congestion as it would for tracking the trailers entering and leaving a concert venue. ‘It can be deployed within 10 minutes, and communicates via mobile telephony,’ says Hewitson.
The system is battery powered, using a cell similar to a car battery, and the control mechanism, camera and illumination are all contained on board. Hewitson attributes the system’s portability in part to the use of smart cameras, which offer image processing in a compact package and low power consumption compared to a computer system.
However, it’s not the ideal solution for everyone. In some situations, traffic monitoring systems are needed day in, day out, year after year, and it would be pointless to adopt a portable system. In these situations, the recent popularity of Gigabit Ethernet communication, which provides fast data transfer over very long distances, has triggered a new trend for security cameras to be connected to a central server at remote locations.
This has two main advantages: it is cheaper to perform all of the image processing from a network of cameras on a central computer, rather than paying for individual processors (or expensive smart cameras) at each security gate; and it is safer to house the processors in a secure, air-conditioned building, rather than exposing the equipment to outdoor conditions.
In addition, sometimes the cameras’ settings need to be altered to suit lighting conditions, and the ethernet connection allows these parameters to be controlled remotely. In some cases, operators could even control the cameras over the web, from anywhere in the world. ‘All in all it’s quite a step forward for how powerful the systems can be,’ says Stuart Box, the director of Cybertronix, which provides these systems.
Mark Williamson, the director of Firstsight Vision, agrees: ‘Gigabit Ethernet is making a big difference for ANPR systems. If I look back to other systems, the cables could not reach far enough away, so we needed to place the intelligence on board the camera. But now, we can distribute the processing away from the cameras.’
Williamson says this is a very efficient use of resources for speed cameras, since the image processing required to recognise the number plate is only necessary when a vehicle has caused an offence, meaning processors on board each camera would be left idle for large periods of time. With a network infrastructure, however, the central server can be kept busy with jobs from many different cameras across different locations.
As with all vision applications, the illumination of ANPR systems is of the utmost importance. According to Williamson, the infrared illuminators used in security systems are not bright enough to give a good image at large distances.
LEDs are an increasingly popular light source, but they dissipate an enormous amount of heat if left running continuously at the required brightness.
Firstsight Vision has solved this problem by synchronising the LEDs with the camera to produce short, but bright pulses of light at the point when the picture is taken. In this way, the time period does not last long enough for the heating effects to build up, producing impressive results. ‘If we pulse the LED illumination we can obtain three or four times the amount of light from them,’ says Williamson. ‘It allows images at further distances, of better quality.’
This consistent improvement in technology is a sentiment that has been echoed throughout this article: transport monitoring systems are proving to be more flexible, more accurate and more intelligent than ever before. It may be a curse for those trying to avoid speeding fines, but ultimately these innovations in technology will provide greater security and safety for everyone.
KEEPING PASSENGER SAFETY ON TRACK
It must be the sentence most hated by every commuter. You’re standing at the station platform, on your way to an urgent meeting, when a train announcer broadcasts over the loudspeaker: ‘The service to.... will be cancelled, due to leaves on the line.’
Hopefully, a new machine vision system recently implemented by Firstsight Vision may help to remedy this situation, by providing an efficient method for train operators to monitor the state of the railway tracks across their network. Previously, operators had performed this manually, which is obviously a very slow and laborious process, but the new system collects the images as the train travels along the track, providing faster and more reliable coverage.
The system consists of nine cameras, mounted underneath the train, which are synchronised with the motion of the train to capture images at regular metre-long intervals. This allows the operators to find the unique geographical position of each image. Once the images are fed into a central database, it is possible to view past records to trace how the track has deteriorated through time.
The underside of a train is clearly a dirty, inhospitable environment for any electrical equipment, so it was essential to create a robust system. The company used military-spec cameras from JAI encased in a rugged package. The team also had to find a way of cleaning away the dirt thrown up from the track. The answer was a cleverly designed unit that directed the airflow to drive dirt away from the cameras, leaving a clear view to take pictures.
All of which helped the system to comply with the rail network’s stringent regulations. ‘One of our biggest costs was putting it through testing,’ says Mark Williamson, director of Firstsight Vision.
No comments:
Post a Comment