4 minute read
License to secure
Ekaterina Mutina, Marketing Manager for TRASSIR looks at how automatic recognition systems for vehicles are helping to keep businesses running smoothly and safely and supporting smart city management
The use of automatic licence plate recognition systems has become more widespread in the Middle East –partially fuelled by the growing trend for implementing smart cities, as well as the advances being made in artificial intelligence technology which is helping to power these systems.
In order to meet the objectives of a smart city, it’s vital that the necessary elements are in place and this includes technology for the precise controlling and management of traffic conditions, increased surveillance and safety on the roads and the avoidance of traffic incidents. Therefore, the deployment of automatic recognition systems for vehicles should be a key priority in smart city management.
Automatic number plate recognition (ANPR) systems have already been used for decades, predominantly by police forces around the world. But their use has now spread to organisations who are looking to introduce their own vehicle surveillance methods as well as monitor parking situations.
Although the use of auto recognition systems for vehicles is a relatively new concept outside of the police force, many businesses already rely heavily on such security systems. After all, they have many advantages to offer securityconscious organisations. The distributed structure allows you to maintain operations even when individual elements of the system break down. In addition, the camera requirements are minimal because processing of the video stream is carried out on the server. The effectiveness of this technology has already been proven and is successfully used for a range of tasks: n Restriction of access to the territory n Organisation of parking n Organisation of payment for parking services n Traffic flow management
There are territories that not everyone is permitted to enter, and in this case, the licence plate recognition system is one of the most convenient and inexpensive ways to restrict access to unauthorised vehicles.
This includes paid parking lots in shopping and business centres, private parking lots designed for storing cars overnight, intercepting parking lots, and many others.
The licence plate recognition system for paid parking is capable not only of identifying incoming and outgoing vehicles, but also of automating the payment process.
In urban conditions, authorised vehicles require passage to a particular territory. These can be cars of special services, such as police, ambulances, the Ministry of Emergency Situations, city service vehicles engaged in street cleaning or rubbish collection, as well as public transport – buses and minibuses, ordinary taxis, or car-sharing vehicles. With the help of licence plate recognition systems, you can flexibly configure access levels and create territories where only certain types of transport are allowed to enter. n Managing vehicle time spent on the territory n Vehicle registration n Vehicle tracking
In many cases, there is a need to limit not only the entry itself, but the time a vehicle spends on the territory. This is in demand at airports, train stations, metro stations, transport hubs, intercept parking lots, and residential territories.
Sometimes it is necessary to simply register all incoming and outgoing vehicles, for example, to collect statistics that allow you to analyse traffic congestion.
Auto recognition systems can also track the appearance of vehicles from a watch list created especially for this purpose and issue an alarm signal when they appear.
Capabilities for security
The market offers different versions of the vehicle recognition system, but as a rule, it is hardware and software based on AI technologies. As a general rule, these systems have a wide range of capabilities that include the most basic – such as automatically detecting and determining the vehicle type, the licence plate in real-time and then storing it and cross referencing it to white/black lists.
In addition, many also offer the option for detailed reports, as well as integration with speed measuring devices and security complexes including access control and re alarm equipment (barriers).
AI for LPR
Note that the degree to which AI technologies are applied plays a great role in the work of vehicle detection systems. Many tasks are solved more e ciently with AI than using standard mechanisms of the past.
For example, in the TRASSIR system, the neural network is able to accurately determine the coordinates of the corners of the licence plate (even if it is located at an angle to the camera) for further alignment, which allows it to determine and record the car’s data as accurately as possible. It has been proven by experience that the system’s implementation allowed users to save money on the maintenance and development of facilities in the elds of construction, retail and industry. In some cases, the time of vehicle control at the checkpoint was reduced by half.
The latest improvements in the eld of auto detection have been aimed at improving vehicle tracking using AI technologies, which are the basis for a more accurate classi cation of cars in the future. Since vehicles can greatly di er visually from one another, the neural network calculates a vector of unique features of the car to improve the accuracy of recognition and tracking. This allows cars to be better and more accurately classi ed by appearance.
Market trends are such that systems are constantly being improved. Manufacturers continuously collect feedback from users, and each example is taken into account when making changes to the end product. We can look at the following areas for improving the module:
■ Training on new data
■ Application of the latest neural network architectures
■ Improvement of the quality monitoring process
■ Other unique developments. www.trassir.com
The six steps of licence plate recognition
As a rule of thumb most licence plate recognition algorithms follow the below steps:
1 Firstly the LPR engine looks to identify the positioning of the licence plate within the image – this is known as framing or localisation. This steps helps the LPR to focus only on the licence plate and disregard any other data.
2 Angular corrections help the LRP to decode licence plates that have been captured at awkward angles –for example on the side or from above. The algorithms also correct for perspective too.
3 Filters are then applied to help eliminate shadows and any shaded areas. Edge detection, in particular, is used when there is a high contrast between the background and the text being identi ed. Some systems use sophisticated technology that blends together multiple images of the same licence plate to get the clearest image.
4 Tools such as whitespace delineation are used to nd the space between letters on the licence plate. Character segmentation works by detecting each individual character. Errors are more likely to occur here if the spacing of the characters on the licence plate is variable.
5 Optical character recognition techniques are then used to identify each character. This could include pattern matching, proportion, pixel repetition and edge tracing.
6 The nal step checks the characters identi ed and their sequence against rules that are speci c to each region.