Goal of the project was to develop a recognition
tool able to provide high coin acceptance and rejection rates.
The tool was integrated in coin validator equipments produced
by IPM.
On AIR main tasks were: |
|
design and development of the classification tool,
operating on the basis of coin parameter sets generated by coin
validator. The classification tool applied neural networks (MLP)
and statistical methods (KNN) |
|
|
analysis of validator sensors outputs with respect
to classification requirements |
|
|
study of alternative classification methods, to
be used on a cooperative basis with the current ones |
|
|
definition of a proper training set, validator
independent, to be used for classification |
