Quantification method of damage state of soiled banknotes

摘要

The present invention relates to a method for quantifying the degree of damage to an old banknote, and more specifically, to perform learning through a convolutional artificial neural network (CNN) on a number of banknote images, and quantify the degree of damage to the banknote based on the learning performed. By extracting the parameters optimized for this, and quantifying the degree of damage to the bill through CNN based on the extracted parameters, it is possible to be provided with an objective value of the degree of damage to the banknote that is difficult for a person to judge objectively. It is easy to grasp the degree as a quantitative value, and it is possible to flexibly set the allowable range for discrimination of banknotes, so it is possible to efficiently set the criteria for discrimination of banknotes of different financial automation devices by country and model. It is about.