Development of Feature Extraction-Based Currency Recognition System Using Artificial Neural Network

Development of paper currency detection system is one kind of smart framework which is of significant requirement in today’s modern world. Currency recognition means denomination classification and counterfeit detection. There are many currencies all over the world. Every currency appears unique based on features such as variation in size, texture, colour etc. In the present study, a system is developed to reduce counterfeit detection time. Image processing and feature extraction technique in the presence of UV light are implemented for currency identification. Some of the features of real banknote are only visible in the presence of UV light. In case of Indian currencies, security thread looks green in presence of UV light. This study trains a feedforward backpropagation neural network with 20 samples of each denomination. The validation resulted 100% accuracy with reduced processing time for currency recognition and counterfeit detection.