Software detection of banknote imagesSteven J. Murdoch

Computer LaboratorySecurity Group

Central banks are increasingly concerned that inexpen-sive home and office equipment can produce passablecounterfeit banknotes. The number of small-scale coun-terfeiting operations have increased accordingly.

Banknotes are carefully designed to show aliasing pat-terns when a low-resolution copy is made. However,this is ineffective against modern high-resolution copy-ing equipment:

Next, a feature was added to colour photocopiers, tostop them printing banknotes if they detect the EURionconstellation, a pattern of small circles that has beenadded to many countries’ banknotes:

More recently, imaging software, such as Jasc Paint ShopPro and Adobe Photoshop, and home inkjet printers andscanners implemented a new system to prevent bank-notes being scanned, or their images being processed:

The way this system works is not publicly known, but bytesting different images, we can learn more about whatfeatures it detects. For example, this figure shows whichcropped areas are detected on their own:

Some minor image manipulations prevent the recogni-tion of currency:

Experimenting with parameter thresholds gives cluesabout the detection algorithms used; as does observationof intermediate computation results, extracted throughdebugging tools:

→

Input Output

→

For more information see http://www.cl.cam.ac.uk/users/sjm217/projects/currency/2006-07-05