Real-time image processing, Pattern recognition, Industrial signal processing

NoteWear: Machbarkeitsstudie zur Verschmutzungsmessung von Banknoten

01.09.2008 bis 31.05.2009

To ensure a smooth supply of banknotes and to safeguard the quality of the banknotes in circulation, some national central banks have reached an agreement with credit institutions that banknotes are to be reissued via ATMs. One of the priority tasks is to detect the quality (soiling, damage, etc.) of the euro banknotes in circulation. For this purpose, the European Central Bank (ECB) has defined a framework for the recycling of euro banknotes.
In order to comply with the guideline of the ECB framework, it is necessary to classify and sort euro banknotes according to their degree of soiling. The challenge here is to classify the soiling of the banknote without a reference note. A method is needed that can generate the reference from the banknote itself.
In this project, a new method for self-referencing feature generation based on statistical features - based on histogram analysis - was investigated. In this procedure, the reference value is generated from the euro banknote itself.
This procedure allows the degree of soiling of euro banknotes to be identified and evaluated. Subsequently, the euro banknotes are classified by gradual assignment to the classes fit (clean) and unfit (soiled).

This project is promoted by:
Employees: Alexander Dicks, M. Sc.
BEB Industrie-Elektronik AG, Schweiz