Cash Currencies Recognition Using k-Nearest Neighbor Classifier

Abstract

The appearance of the currency is part of this development and it is affected directly, where there is exploited in incorrect form by copying the currency in a manner similar to the reality. Therefore, it became necessary to implement a proposal for being a suitable as solution not inconsistent with the different cultures, time and place, to reduce the risk of problem that represented in distinguish between real and fake currency. This clear through add the watermarks inside currency, which is difficult to be copied. At the same time, this watermarks may be visible to the naked eye so can easily inferred or it is invisible. However the high resolution imaging devices can copy these additions. In this research, we have proposed a system to distinguish the currencies by the program that working a submission inferred to the watermark by feature extraction determined the type of currency and its reality. In addition to, the algorithm (k-NN) determined category of the currency. Benefit of it, is reducing as much as possible the spread of counterfeit currency and this system can be used by any user wants to make sure of the currency reality. The proposed model applied on 100 banknote, the success rate was 91% and the failure rate was 9%.