Recognition of car numbers: foreign experience..
Much has been said and written about video analytics systems designed to recognize license plates. Enormous public attention in Russia was drawn to the recently introduced practice of using a license plate recognition system and imposing fines for this or that violation of the Traffic Regulations based on the data from this system. The market for protecting license plates from recognition responded, as always, sensitively and promptly, offering the consumer a wide range of miracle cures. As usually happens when introducing new systems and practices, there are more questions about the legal legitimacy, technical reliability, and ways to combat this system than there are answers.
In order to try to give objective answers to such questions, we will analyze foreign experience in the development and use of video surveillance systems that allow recognizing license plates. Usually, the Russian market for video surveillance systems repeats, with some modifications, the main stages that foreign colleagues have gone through. Therefore, let us first consider how license plate recognition systems have been used abroad over the past few years.
IMS Research analyzed the situation in the USA in 2010 and found that the total volume of the current license plate recognition market is million [1]. According to the company's forecasts, this amount will approximately double in the next five years. Such high volumes of this market segment indicate that license plate recognition systems have proven themselves to be efficient systems that are used in many areas. The same study also reports an expectation of an increase in the international market for such solutions. In the future, significant funds are expected to be allocated for the purchase of such products.
Thus, in 2011, a modern and high-tech video surveillance system was installed in India, in the Maharashtra police academy. This happened after an agent of a Pakistani terrorist organization was detained with photographs of this academy. The designer and installer of the solution was the local company Revmax. The system is built on the hardware base of the Indian-American company Cradle Technologies [2]. The cameras «cover» over 100 hectares of area. This video surveillance system, along with the functions of identifying faces, including in makeup, transmits acoustic information in two directions, including address and mass messages. In addition, this system reads and registers car numbers.
The effectiveness of video surveillance systems that recognize license plates is demonstrated by the methods used by Chinese violators to combat these systems. In 2008, Chinese «masters» launched the production of devices that can change license plates on a car in a matter of seconds. For example, the police in Guangdong Province state that many photos taken by surveillance cameras show license plates covered with either opaque material or fake license plates [3]. The cheapest model of the device costs approximately 800 yuan (less than 3,000 rubles). A more advanced version, which costs twice as much, can change a license plate in less than three seconds. In other words, attempts to deceive modern video surveillance systems that recognize license plates by significantly exceeding the speed limit, creating additional interference by applying varnish, mesh, or additional backlighting to the license plate have not been successful. Otherwise, purely mechanical means of installing a new license plate would not have found such wide application.
Goals and objectives of the vehicle number plate recognition system
Back in 2005–2006, when building video surveillance systems abroad, much attention was paid to recognizing vehicle number plates.
For example, in 2005, the Philadelphia school district required video cameras to take pictures of license plates located within a block and a half of schools. A solution was developed in which the cameras were mounted on remote-controlled swivel units. Each of these cameras cost approximately $1,000 [5].
In 2006, 150 video cameras were installed in the city centre in response to a request from the Stockholm municipality to read number plates. They are used to monitor traffic flows, measure their density and monitor traffic jams. The system includes video cameras capable of reading number plates. They integrate with standard digital video recorders and are capable of performing their functions both day and night, in complete darkness, at high vehicle speeds and in the blinding light of headlights. All products in the line are integrated with automatic number plate reading (ANPR) software. This integration allows for a quick search of numbers in databases [6].
The Birmingham-East Midlands motorway is equipped with high-quality CCTV cameras with focal lengths of 7.5; 16; 50; 75 mm. The images are transmitted to the Highways Agency's regional control centre in Quinton and recorded digitally. It is envisaged that at times of greatest congestion on the M42, when traffic jams occur, two additional lanes on the hard shoulder of the motorway will be opened for traffic. Each time before opening the additional lanes, they will be inspected by the system operator using CCTV cameras to detect possible obstacles to traffic, such as abandoned cars or rubbish [7].
It is noteworthy that license plate recognition systems abroad are used not only and not so much for issuing fines to traffic violators, but also for searching for persons committing any crimes and located near socially significant objects. Due to the ability to quickly search in databases of recognized license plates, these systems cope with the analysis of operational information received during patrols. That is, license plate recognition systems, due to the widest integration with various databases and software that allows analyzing the results of processing this data, can offer various optimal algorithms for the behavior of security structures in connection with a particular situation. In addition, these systems are successfully used to combat traffic jams; they help analyze situations on the roads.
In 2006, Virage developed a highly advanced system for recognizing and processing vehicle license plates [8]. Because this license plate recognition system can be linked to multiple archives, such as police and traffic flow data, it can quickly provide information on the relationship between registered license plates and other characteristics of any given vehicle. For example, if a vehicle is detected approaching a high-security area (such as a child care facility or government building), the Virage CCTV system will automatically alert personnel to the fact that this is the fourth time the vehicle has passed through the area in a given period of time. In addition, the system will automatically initiate programmed responses once the vehicle has been identified. For example, if an unidentified vehicle is detected, the access control system will deny access to the vehicle and, if necessary, alert the CCTV operator.
In the British city of Bath, 46 video cameras have been installed to monitor the main roads entering and leaving the city. They are connected to a computer system that can read the number plate of each car passing by and compare it with a database [9]. This number plate recognition system uses 30 databases, including national ones, held by the police. When the system determines by the number of a car that it belongs to a wanted criminal, the nearest patrol can intercept such a car. In addition, the system makes it possible to catch not only wanted criminals, but also people who drive a vehicle without a driver's license. For this purpose, it is connected to the system of the administration for issuing driver's licenses and motor vehicle licenses.
GE Security has integrated Survision's Autofocus software into the VisioWave digital intelligent video surveillance platform to enable automated parking access control [10]. VisioWave users using the system for police, bus, tram and passenger car services can identify vehicles by license plate number to manage and control parking access. The system can also identify stolen vehicles and automatically record parking time for cars, trucks and trailers.
Another interesting document is the results of the CitySync ANPR pilot programme [11], published by the UK Home Office. CitySync A is widely used not only in city CCTV but also in police patrol cars. CitySync ANPR systems have been installed in UK cities such as Aylesbury, Canterbury, Liverpool, Portsmouth and Windsor. The system can operate at a wide range of speeds, even exceeding 100 mph. Infrared cameras capture images in low light, rain, reflections and glare from headlights. Software reads number plates and matches them to databases, alerting officers within seconds if stolen, unregistered, uninsured or taxed vehicles are detected. Number plate data is extracted and matched against 60 police databases in real time. When the search parameters and the information in the database match, an alarm signal is triggered by the system controllers, which initiate the appropriate actions. CitySync also installed its license plate recognition system in 2005 at one of the most modern power stations in the UK, Cottam [12]. The 2 GW thermal power station is located 13 km from the village of Retford in Nottinghamshire. The Eclipse Digital Software software built into this system allows for the integration of additional software modules, both proprietary and from third-party developers. The latter circumstance was of particular importance, since the Geutebruck Multiscope System had already been successfully operating at the station. Video cameras were installed so that the license plates of cars stopping at the checkpoint could be identified. All persons entitled to enter and drive onto the station’s territory were entered into the database connected to the system. When an unknown vehicle was detected, an alarm signal was transmitted via the Geutebruck Multiscope System to the central monitoring system.
In 2006, the British company Appian Technology installed video systems for recognizing license plates on patrol cars, winning a large order for £300,000 [13]. This was an additional order from the British police department. According to experts, such systems installed on cars are highly effective in operations to capture dangerous criminals involved in illegal arms and drug trafficking.
Sources
1. http://imsresearch, 12.2010
2. http://cradle, 02.2011
3. http://imsresearch, 10.2008
4. http//www/ingcom, 12.2005
5. http://derwent, 08.08.2006
6. http://rainbowcctv, 01.21.2006
7. http://triangel.sk, 06.29.2006
8. http://virage, 04.2006
9. http://bewator, 01.2006
10. http://gesecurity, 05.2006
11. http://citysync.co.uk, 01.2005
12. http://citysync.co.uk, 08.2005
13. http://marketwire, 02.2006
(To be completed in the next issue)