The banknote manufacturing industry is shrouded in secrecy, the fundamental mechanics of security components are closely guarded trade secrets. Currency forensics is the application of systematic methods to determine authenticity of questioned currency. However, forensic analysis is a difficult task requiring specially trained examiners, the most important challenge is systematically and methodologically repeating the analysis process to reduce human error and time. In this thesis, an empirical approach for automated currency forensics is formulated and a prototype is developed. A two-part feature vector is defined comprised of colour features and texture features. Finally the note in question is classified by a Feedforward Neural Network (FNN) and a measurement of similarity between template and suspect note is output. Colorspace performance have been compared namely the: RGB, HSI, and Lab colorspaces. It is found that the combined average between the RGB channels known as the Intensity channel provides the highest discriminability, and is selected as the candidate colorspace.

By its definition the word currency refers to an agreed medium for exchange, a nation’s currency is the formal medium enforced by the elected governing entity. Forensic science is the application of scientific methods to answer questions of a legal nature. Throughout history, issuers have faced one common threat, the threat of counterfeit. Despite technological advancements, overcoming counterfeit production remains a distant reality. Scientific determination of authenticity requires a deep understanding of the raw materials, and manufacturing processes involved. This thesis serves as a synthesis of the current literature to understand the technology and the mechanics involved in currency manufacture and security, whilst identifying gaps in the current literature. Ultimately, a robust currency is desired, a robust currency is one which withstands security breaches, and is durable surpassing the lifetime of the current currency.

It has been identified that the current currency forensic investigation process is a manual ad-hoc process requiring specialist sought after questioned document examiners (QDEs), clearly this process is subject to human error. In a forensic setting, the analysis process must be systematic, methodological and repeatable. The digital currency forensics system addresses the issue of currency analysis by implementing a specific repeatable process through an automated examination using a combination of image processing, and classification techniques. This is achieved by implementing machine learning and pattern recognition.