Authenticity Assessment of Cuban Banknotes by Combining Deep Learning and Image Processing Techniques

The rapid evolution of technology has significantly increased the production and circulation of counterfeit currency, particularly high denomination bills, posing a detrimental impact on society’s commercial and economic sectors. The continuous advancement of domestic technology makes it increasingly challenging to differentiate between genuine and counterfeit printed money. Despite the development of computational tools for detecting counterfeit currency in various countries, no such tools currently exist for Cuban banknotes. This study proposes methods that combines deep learning techniques and other classifiers based on shape, color, and texture features applied to regions of interest, as determined by forensic experts who consider their security measures and characteristics. The evaluation of the proposed methods shows their effectiveness in detecting counterfeit currency, even with the limited availability of both fake and genuine specimens.