RecommendMail Facebook Twitter LinkedIn

Automatic license plate recognition system – a system that learns

A system inspired by biological neural networks that ensures more safety and comfort and contributes to protecting the climate.

Many people may still remember license plate guessing games on long trips. Who knows which city the letters stand for on the license plate? The more you play the game, the better you are at it. That's because we've learned to assign the license plate to the right city while training our brain at the same time.

Our engineers have also worked intensively on the subject of license plate recognition and developed an algorithm that learns in just the same way as our human brain – perhaps even a bit better. Artificial neural networks reproduce the way in which the human brain functions. So-called "deep learning" is used in our automatic license plate recognition system (in the UK: automatic number plate recognition, ANPR).

The basis is an imaging process in combination with a "deep learning" algorithm. It learns from a huge dataset of sample license plate images. The system only learns during the training phase in our labs, where our engineers give the system feedback if its answers are correct or wrong. The learned network achieves 98 percent recognition accuracy, which compared to conventional methods, represents an increase of 2 percent. This may not sound very much but in a safety or revenue critical system, it could be the difference between success and failure.

It also learns which license plate is assigned to which country, along with the class and color of the associated vehicle. Additionally, our "TraffiData" analysis software also makes it possible for correlations of the stored data to be made visible in real time, such as automatically identifying movement profiles and behavior patterns for suspicious vehicles.

Increased road safety

With our solutions, we support police and security authorities to increase road safety worldwide. The world population is growing year by year and cities and their infrastructure are in pace with it. Around 7.7 billion people live on our Earth today and by 2030 there will be 43 mega-cities with more than 10 million inhabitants.

This will also increase requirements on mobility, security and traffic monitoring. Authorities need effective and reliable measuring devices that can check vehicles in real time. Alongside increased safety in the flow of traffic, our system also helps enable intelligent traffic route planning and avoids traffic hotspots. Our ALPR (or ANPR) solution can be customized depending on the purpose.

At the same time, automatic license plate recognition also offers parking lot operators more convenience. Manually checking the well-tried parking disc is becoming a thing of the past, thanks to optical systems at the entrance to parking lots. They automatically detect when the parking time of a vehicle has been exceeded. Similarly, placing automatic license plate recognition systems at fast-food restaurant drive-ins would provide the option of offering personalized order suggestions.

Customizable solution – ALPR

To protect the climate, methods also need to be implemented and regulations and requirements controlled.

For example, driving controls due to high levels of gaseous emissions or particulate matter in cities can be controlled efficiently, with minimal infrastructure and staff costs. Only a technical solution like ALPR can make this possible.

Safety is our top priority.

Kevin Chevis, Head of Light & Safety