Technologies of license plate recognition are quite common thing lately. Moreover, current security requirements make plate recognition very optative. Installation of recognition systems allows to control presents and movement of vehicles on the territory of different sizes and the creation of the video archive allows to find offenders quickly and effectively. Integration of the system with parking billing minimizes influence of human factor and makes impossible the employers abuse of authority. Integration with video fixation complexes of traffic regulations provides secure control of highways: identification of offenders by the license plate allows automatic preparation of penalty tickets and also to detect the vehicle in the race which can concern law enforcement agencies.
In fact, automatic recognition of license plate is the ability of the system to allocate characters from image automatically and then manipulate with it. Substantially, automatic recognition consist of “frame capture mechanism” which has the ability to fix the image, search for location of the plate on image and following allocation of characters with the help of optical character recognition (OCR) transforming pixels into readable numbers.
There are two ways of such fixation.
First one is OCR of every letter/number of the whole license plate and provision of approximate values. This way is simple and chip, but it cannot ensure reliability of results: chaotic optical character recognition doesn’t take into account an inspection angle and slope and also can be confused by background events and therefore provides an inaccurate recognition.
Second one is more compound way and consists of usage of neural network which allows sequentially set up every plate and provides more reliable recognition. At first plate’s data defines and then applies the rules of partition the plate on alphanumeric components, thereby eliminating the background needlessness, created by marks of distinction provinces, images and other details, which can be seen on the plate. The neural network can provide recognition accuracy nearly 97% in all conditions. With realization of those mechanisms developers should solve some problems such as blurring of characters of license plate as the result of bad focusing or driving, sharp angle images due to poor camera angle and low contrast. Also, plates may turn out blurry because of bright light of the headlights or because of reflection of the sun lights, and in addition, a lot of other factors may effect on the clarity of the image. In general, in spite of wide range of application and fairly active temp of development in this sphere, developers have to solve many problems and the quality of the system depends on the positive outcome.
Long-standing experience of our company in this sphere allows us to find the solution of most of these problems and become one of the leaders on the recognition systems market. We offer the solutions for big companies and governmental structures and also for owners of small cars parks and petrol stations. Our developments on the field of automatic license plate recognition are:
Overseer Parking systems was developed for optimization and automation of parking lots and parking facilities. It allows to build an integrated system of events control on the parking lots, keeps customers records binding to their plates and flexibly customaries the events which should take place in different conditions on predefined rules. During Overseer Parking system implementation one administrator can maintain parking for hundreds of vehicles with many inputs and outputs without leaving his office.
Integrated customers base is connected with account balance and parking automats at the inputs that allows clients do almost all items of interaction with parking by themselves.
Overseer Traffic is oriented on governmental structures primarily. The feature of this system consists in the possibility of external devices connection and their operation according to flexibly definable rules. For instance, using LED signs on the bridges and columns allows the possibility of automatic creation of the corridors for vehicles with special purposes, optimization of the speed of traffic lights turn on and off in dependence of intensity of the vehicles flow. Also, Overseer Traffic allows automatize the creation of protocols and traffic violations and penalties statements.
In many cases, license plate recognition systems are needed not only for stationary but also for mobile objects. An example is the security service, car security and traffic police, special cars. Systems, like the Overseer Mobile, are indispensable for carrying out of search operations, investigations of carjacking in hot pursuit and search of specific vehicles during counterterrorism operations. Herewith the algorithms of image processing which are built into the Overseer Mobile, automatically compensate wide angle distortion. As a result, quality and capture license plate recognition is reduced by less than 5% compared with stationary versions.
AutoCode plate recognition module is designed for installation in CCTV "Intellect" (ITV, Russia), and XProtect Basis +, XProtect Professional, XProtect Enterprise and XProtect Corporate (Milestone Systems, Denmark).
