Your number is up, says Institute of Systems Science to traffic offenders

CATCHING motorists who break traffic rules will be easier with new technology that can read vehicle licence plates automatically from pictures, such as those shot by speed and red-light cameras.

The technology developed by the Institute of Systems Science (ISS) is able to feed the numbers to a central database within seconds of the images being captured on camera.

A live video feed can be used to transmit images to a central computer. Sensors triggered by a speeding vehicle, for example, will send digitised pictures of the vehicle to the computer.

The computer zeroes in on the licence plate and feeds its number into its database.

Dr Tian Qi, 51, a member of the research team, said that the system's uniqueness lies in its ability to decipher licence plate numbers even if the plate's surface is scratched, slanted or smeared.

"We can also program the system to read the licence plates under poor lighting or poor road conditions," he said.

Infra-red strobes rather than conventional ones could be used in poor light.

"Motorists will not know that pictures of their cars are being taken," he said.

This is an improvement over the existing cameras which require film to be collected and processed manually. It took the team more than five years to develop the latest identification (ID) technology.

With motor cars, trucks and vans, the system identifies the right registration plate 95 per cent of the time. Accuracy drops to 50 to 60 per cent for motorcycles and scooters but Dr Tian expects this to improve.

"It is more difficult for the computer to process motorcycle licence plates because of the different ways the plates are positioned on motorcycles," he said.

Catching traffic offenders is only one of many possible applications, he added.

"It could be used in traffic control management like car-park management and car tolling systems, for example, the electronic road-pricing system," he said.

Carpark operators could do away with conventional ticketing systems. They could use cameras to record licence plate numbers when vehicles enter and leave a carpark.

Since 1993, the Port of Singapore Authority has been using 19 of these ID recognition systems to check the identification numbers of containers passing through its gates.

It is working with the ISS to develop another customised system to check the containers being loaded and unloaded.

Dr Tian said that installation costs depend on what the system is used for. A carpark operator, he said, would have to spend between $10,000 and $30,000 to install a system.

Another team member, Mr Goh Tiong Hwee, 38, said that ISS would set up a spin-off company to market the ID recognition technology.

The company will be operational early next year. It will approach potential customers such as the Land Transport Authority, Singapore Bus Service, Urban Redevelopment Authority and other carpark operators.