Today, over 80% of Dutch people own a smartphone. A recent study indicates that people use
their phones more than 200 times a day for a total of more than three hours. They use it for almost anything and carry it with them day and night. Now, after several years of research by
De Nederlandsche Bank (DNB), the smartphone can also be used for banknote authentication.
This article will present DNB’s new ‘Genuine or Counterfeit’ app, and will shed some light on the development process and how this app fits in DNB’s communication policy.
How did we start
In 2011, DNB launched its first smartphone application. This app called ‘Eurobiljet’ was aimed at educating the Dutch public on the use of the security features of euro banknotes. The app shows a virtual banknote. When you hold your smartphone up, as if holding a banknote against the light, the virtual banknote on your screen shows the watermark. In the same way, tilting your device allows you to explore all the images of the holographic foil.
Soon after the introduction of this app, several users asked us whether we could make an app that could scan and authenticate the banknote by itself. This was easier said than done, but we started working with this idea, and in 2015 DNB launched its ‘Genuine or Counterfeit’ app that can do just that.
In 2013 we started a research project aimed at using the smartphone camera for banknote authentication. One of our requirements was that the solution should work for all existing euro banknotes currently in circulation, as we did not want to wait for an introduction of new notes or new features on the note. Together with several partners, we worked out various ideas, such as taking a photo of the watermark while holding it against the light or looking at the red to infrared parts of the photos. We realised that the best option would be to look at the printed information on the banknote as a whole. We joined forces with Perceptech BV, a spin-off company of the University of Amsterdam with expert knowledge on image recognition and computer vision, to further investigate this option.The print on euro banknotes is rich with complicated and detailed structures. Although counterfeits can be made that look quite similar, it is difficult to copy them exactly.
This is mostly due to the fact that reproduction equipment used by counterfeiters can differ significantly from the line screen printing equipment used in the security industry. As a central bank, we had one major advantage during this study: we had access to all discovered counterfeited banknotes. Scanned images of a large number of genuine and counterfeit EUR 50 banknotes enabled us to investigate the differences between these two groups.
A precondition for comparing the images was finding the correct position and dimension of the notes. For this we used key-point detection, followed by a correction of the image distortion. This produces a perfectly rectangular image of the note with a predefined size. Printed elements can then be found and investigated. Fine details of the banknote print are consequently compared with the image of a genuine note, to deter-mine whether the note is authentic. A feasibility study with 1000 counterfeit notes and 700 genuine notes from circulation showed that 99.4% of the genuine notes were classified correctly, while none of the counterfeit notes were classified as genuine. The 0.6% mis-classification of genuine notes was due to wear and tear of these notes.
After the successful feasibility test we decided to develop a prototype that would run on smartphones. By the end of 2013 we had finished a concept version of the app for the iPhone 5. The app worked well with EUR 50 banknotes, but the processing time was time consuming. Luckily, at the same time, the iPhone 5s was launched, with a stronger processor and an improved camera. We obtained one after some queuing at the Apple store. We found that our software was running faster on the new iPhone 5s and processing times would be more acceptable for the general public.
In order to be able to include all denominations in the app, we collected information on genuine notes and their general quality and built the banknote authentication model. This model was then validated by testing it on all counterfeit types registered in the Netherlands. Retesting was done to consolidate the final version of the app for the latest iPhone models, a time and resource consuming process. In May 2015 the app for the iPhone 5 and newer models was finally released.
How does it work
The app user interface was designed by Visualspace, a media and technology company. As the app had to be both fast and intuitive, it took several rounds of testing to find the optimal user interface.The app only has one function: to check whether a banknote is authentic. The opening screen is the viewfinder of the camera. By swiping, the relevant euro denomination can be selected. A template enables the user to position the phone correctly over the note. After this, the app will recognise the banknote, a countdown starts and a photo is taken automatically. This photo is then analysed, and the result is displayed. See below for screenshots of this sequence.
The app will recognise genuine notes. If there is not enough resemblance, the banknote is not recognised as genuine. In such cases the app does not state that the banknote is counterfeit, because it could also be crumpled or worn but still genuine. If the visual infoRmation is partly lost because of wear, the note is not recognised. The user is then informed that the app cannot verify that the note is OK. After the second failed attempt, the user is asked to check the note himself as shown above. In case the note is recognised as genuine, the user is also invited to check the note himself. This underlines our view that the app is an aid, and that it remains important to check the note yourself.
The authentication app has been tested, with positive outcome, on all counterfeit types active in the Netherlands in the past two years – a total of 550 different types. As counterfeiters will keep on producing new notes, the monitoring of the correct functioning of the app is therefore a continuous effort. New counterfeit types are tested as soon as they are registered. After each update, the app is also tested on a representative set of used genuine notes. We will also need to keep updating the app as devices and their operating systems evolve and new ones are released.
The app and DNB’s policy
How do we see the use of our app? DNB stresses the importance of checking banknotes. For shopkeepers, suitable authentication aids have always been available, but for the general public this was not the case. This has now changed, as our app provides a helpful authentication aid for all smartphone users. In the first week of it’s release the iPhone app was down-loaded 50,000 times, which shows that the app is fulfilling a need, and that it stands out between the 1.2 million apps that are available in the app store. People who have downloaded the app can now just take their phones out of their pockets and use it, should they ever have doubts about a note they receive.
An important additional benefit is that the app strengthens our education campaign on banknote features using new media. DNB’s app will generate extra interest in and knowledge on the banknote and its security features. Furthermore, when the result of the authentication is shown, a link to our educational ‘Eurobiljet’ app is provided, so our digital tools are supporting each other. By creating the app we have not only reached the part of the Dutch public that owns an iPhone 5 or a later model, but also the general public as the app was extensively covered in traditional media, such as news-papers, radio and TV. This helped educate even more people on the security features of the euro banknotes.
The app is meant for the general public. Shopkeepers and other professionals should keep using dedicated authentication equipment. These tools are quicker to use and also cheaper than smartphones. For all other users, the app could be a nice aid for those moments when they are in doubt or looking for some extra reassurance. The app is available free of charge in the Apple app store, and at the time of writing, an Android version is being developed.
A public task
Providing information on the euro banknote and its security features is one of the key tasks of the Dutch Central Bank. Supplying our apps to the general public for free is certainly part of this. Of course commercial initiatives that do not involve central banks are always possible, and in fact several apps that claim to detect counterfeit euro notes are available. These apps check the serial number of the note and as some counter-feiters do indeed print wrong numbers on their notes, these apps could occasionally be correct. Nevertheless, it will be obvious that these apps are misleading.
We believe that an authentication app that can be trusted should be based on all known counterfeit types and on the latest examples. Developing such an app should remain a public task for the following reasons:
• It would be questionable if we left the development of this kind of apps to a third party.
• Testing and monitoring the performance of an authentication app requires access to all existing counterfeits.
• After the development stage, monitoring should continue as the counterfeiters’ techniques will evolve. In order to safeguard reliability, access to new counterfeits is needed as soon as they appear. Central banks and police forces are usually the only parties that can provide this.
Given the rapid evolution of smartphones, it is difficult to know what DNB’s app will look like in the future. Not all aspects are in our control. Besides the question whether smartphones will still exist in their current form, even today it is a fact that companies such as Apple and Google control the software on these devices, and app developers such as DNB will have to adapt to that. As for the counterfeiting technology, developments here are unpredictable, and this is why the quality of counterfeits in relation to our authentication technology has to be monitored constantly.
The focus of our future developments will be on improving the user experience. The possibilities for correctly authenticating soiled notes will be improved, and methods will be developed for providing detailed feedback on the reasons why a photo did not deliver a good result. This will allow users to correct the error (for example, the lighting condition) and try again.
Another interesting option is to improve the strength of the current banknote design for smartphone authenticcation. Our app demonstrates that the printed informaction on the banknote is already a good authentication feature. Nevertheless, it can be enhanced, by finding factors that further increase the differences in reproduction techniques of genuine versus counterfeit notes. Here the strategy should be to find enhancements that can cover large areas of the note in order to increase redundancy. Enhancing existing elements instead of introducing new ones further allows a smooth transition from the current series to a new series.
Introducing one dedicated feature to be printed some-where on the banknote is the alternative approach. Again, features that can cover large areas of the note would be preferred. Whatever the feature proposed, one drawback to this solution is that, depending on your issue strategy, it could take considerable time before most banknotes in circulation would be ready for smartphone authentication. If we take the euro banknotes as an example, it would take 10 years or even longer before the large majority of circulating notes would contain a new functionality. Given the rapid developments in devices, apps and entire principles of communication technology, this horizon seems much too long.
One of the key tasks for a central bank is to educate the public on the use of security features. With the ‘Genuine or Counterfeit’ app we keep abreast of the times and provide help and information on a device that most of us nowadays have access to. We believe it is essential to be present on this medium, while meeting the high expectations of today’s public. DNB’s app can fullfil this role and provide a very welcome banknote authentication aid for the general public.
Keeping the app updated, monitoring counterfeits and making sure that the app meets our high standards for reliability does require significant effort. But we believe it is worth it, and one simple but convincing argument comes from a director of De Nederlandsche Bank. After a presentation of the first concept he said, rather satisfied: “This is a very cool app, and it’s ours!”
1 MacNaught, S. (2014). Tecmark survey finds average user picks up their smartphone 221 times a day.http://www.tecmark.co.uk/smartphone-usage-data-uk-2014/
Marcel van der Woude is a Senior Policy Advisor at the Cash Policy Department of the Central Bank of the Netherlands. Presently he manages a diverse range of cash related projects, such as the introduction of sustainable cotton in banknote paper, research into the use of cash, and purchasing of banknotes. Recently, he developed a smartphone app for public education on banknotes. Within the European System of Central Banks he contributes to the committee on quality management of Euro banknotes.
In 1997, Marcel graduated with a Professional Doctorate in Engineering in Chemical Process Design at the Technical University of Eindhoven. Prior to that he completed a Master of Science degree in Chemistry at the University of Nijmegen.