ANPR: Are you getting value?
May 2013, CCTV, Surveillance & Remote Monitoring
Automatic number plate recognition (ANPR) as a technology has been around for a while and a number of suppliers provide products to the industry. However, it has not really taken off in a big way and one wonders about the reasons for this.
ANPR systems are not inexpensive, but various low cost options are available, depending on your need. The cost is affected by a number of reasons. The development of recognition engines is time consuming and expensive and if one needs recognition under adverse conditions this development costs even more. The cost of true ANPR cameras is high. This is partly due to the fact that the volumes are smaller but also due to the fact that a proper ANPR camera consists of more than just a standard camera. To name but a few of the real required characteristics of a true ANPR camera.
1. It must be able to grab a sharp image of a moving object.
2. High shutter speed comes at a cost – there is less light to illuminate the image sensor.
3. A constant source of light wastes a lot of energy and suitable solutions should be found.
4. They use limited spectrum light such as infrared because the IR filter excludes other light sources such as head and tail lights, a large proportion of a glaring sun and reflections, and so on.
Undoubtedly there are systems with poor recognition capability and others with excellent capability. Some work well under good conditions but fail when conditions deteriorate, such as sun reflection, darkness, against headlights, etc. Systems with good image enhancement and then good character recognition behind it offer considerable improvements.
Ignoring the value of ANPR
Perhaps the main culprit is the inability to understand the benefit of having the vehicle number information as opposed to just having recognised it. Many systems simply read the number and overlay it on an image or store it on a video recorder. But if that information is captured in a database a wealth of benefits appear. One can quickly search for a vehicle by time, or number, one can group vehicles together, understand who is in and who is out, authenticate a driver, prevent access and so on. The following applications can be implemented.
Logging entry/exit: Logged data makes it possible to determine whether and when a vehicle was on the premises. In this way one can keep track of the presence or absence of a vehicle.
Managing who is in and out: Scanning the vehicle register can quickly determine whether a vehicle is on the yard, whether its owner is on site or away. A simple filter can immediately say how many vehicles are on site and how many have left. One can now also determine the duration of stay on a site.
Managing content: Linking weighing stations with ANPR systems can log vehicle data on entry and vehicle data on exit – knowing exactly weight delivered or loaded.
Verifying who is abusing a system: In fleet management applications it often happens that a vehicle is refuelled, leaves the premises, transfers his fuel to another vehicle and returns for another refuelling. By monitoring vehicles, dates, time of refill and so on this abuse can easily be stopped.
Controlling access through black and white lists: By using the system to verify the access level of specific vehicles, those which were black-listed because of previous bad behaviour can be refused entry or can be watched carefully. By the same token frequent users can be treated specially by opening the gates automatically and a friendly welcome message can be displayed. This could assist with making customers feel welcome and improve their attitude towards a centre or facility.
Improving customer relations
In addition to the previous, vehicles’ plates can be used to determine demographic information by matching vehicle types to specific areas.
Knowing the identity of a client and when he arrives can allow a vendor to present suitable information to him upon arrival, based on previous contact. Frequent visitors to a shopping centre can be treated to occasional benefits. How about providing a special convenient parking area to frequent visitors – surely it will entice them to prefer this centre over another? Specials or discounts on vehicles that visit a drive-through facility regularly may encourage sales.
But what if we start sharing knowledge between sites and between one another and start to create mutually beneficial applications?
Creating area-wide black lists
Vehicles identified as suspicious in one shopping centre can raise alertness at all the branches of the chain, or, if data is more widely shared at all public areas. Stolen vehicles can be recognised at every facility with ANPR and the police services can be alerted. What if all systems perform a number/colour match and have a centralised repository for mismatches? Stolen number plates can likewise be identified automatically and relevant authorities be notified.
The most obvious integration of data is a link to the e-Natis database for vehicle ownership information, to the police database for stolen vehicle information. While this may be useful, it is unlikely that these databases/lists will be made public. But organisations may build their own databases with information and share these with one another or within their own branches. For example, what if the number plate of a vehicle is automatically linked to the card that was used to pay for the fuel – surely this will disallow the use of stolen cards for fuel payments.
Making the above possible
When developing turnkey solutions for any client that wants to use the intelligence of ANPR systems, local development is important. Not only to customise the ANPR system, but also to assist with the development of interfaces and data presentation to the intelligence users or developers.
Having the results in a stored list is a necessity, but having it in a database makes access, filtering and management significantly easier. If the database is distributed, possibly stored in the cloud, this could make it even more accessible. Of course the ANPR system must provide high quality detection and recognition or else clients will be unhappy.
In many of the above cases there is some need to adapt/design a suitable solution but maybe the era of one size fits all has been so limiting that we have become satisfied with partial solutions. Clearly it is possible to create solutions that work for every individual application and imagine where we can be, if we stop accepting that we are all the same and our needs are all the same. While the ability to develop such solutions may not exist with all installers there certainly are companies very able to assist in building your own solution.
The main point of this article is that perhaps it is time that we start to use the information we gather as intelligence to assist with improving customer relations, actually reducing crime that involves vehicles and so on. As mentioned in other articles in this series it surely is time, and it is definitely possible, to change security systems into information management systems.
Dr Coetzer can be contacted at Protoclea Advanced Image Engineering ( firstname.lastname@example.org, www.protoclea.com or www.facebook.com/protoclea), a company specialising in the development of products such as those described in this article.