BTS LPR

For Norwegian company BT Signaal, we have developed the BTS LPR Automatic Number Plate Recognition system, supported by several OCR engines. It’s a response for growing needs of the transport infrastructure sector associated with traffic management, vehicle location and toll collection.

lpr-fin

The Challenge

How to deliver flexible quality and scalability of the automatic license plate recognition systems while at the same time maintaining the overall quality of recognition and classification?

Solutions to the problem

With a view to the most efficient our client’s system performance, we used OCR engines from different vendors which have been integrated into one system. We have utilised cloud computing in support of this solution, which provides enormous computing power and incredible speed of data processing.

How does it work?

The LPR system, developed for BT Signaal, performs advanced image processing and analysis in the cloud based on several OCR engines that recognize the characters in graphical form. LPR evaluates results for entered images given by OCR engines and chooses the best match. The results of this process are information on the license plate, the country code and the level of confidence. The service also gives us the ability to use other OCR engines and to define other principles for the assessment of results.

In the future users will also be able to utilise this system to recognise both the model and colour of the car. Moreover, system will be supported video stream handling, including detection of vehicles in motion directly from video.

lpr2

Massive computing power

Because of the scalable system architecture, it is possible to process a large number of images. The current capacity is 1 million images per hour.

Scalability

Cloud-type infrastructure allows you to scale up or down depending on demand during the day. This means that the performance of the system adjusts dynamically to changes in the load.

Advantages and benefits

  • Easy and intuitive operation
  • Smooth flow of vehicles
  • Increased economic efficiency
  • Lower operating costs
  • No need for management of infrastructure (functions within the cloud)
  • Scalability of the solution
  • Transaction-based processing
  • Camera configuration support
  • Camera independent, immediate integration
  • Utilization of camera integrated OCR engines
  • Statistical data to improve quality
  • Region-based country distribution
  • Vehicle make and model support
  • License plate type support (taxi, diplomat, personal, etc.)
  • Platform-independent API for accessing the service
  • Optional return of result candidates
  • Optional vehicle registry lookups
  • Optional manual recognition

Applied technologies

netmvcjquerysql2008sasswcfdal21