Banknote Recognition System for Visually Impaired People

Abstract

Paper currency recognition is one of the most important needs for visually impaired people to deal with different life circumstances. In this paper a layered framework is developed in which an Egyptian banknote recognition system is given. The developed system can recognize banknote in all geometric transformation (scaling, rotation, translation, shearing) and all lightning effects .The system is validated by using dataset which contains 1200 images. The underlying image dataset has been divided into five main categories (clear , rotated, noisy, folded and old images).The system is tested and the recognition accuracy is 93% in all categories. The first layer in the framework deals with watermark security. The second layer of the framework, the segmentation has been applied, so that its value, special textures and special drawings are extracted. In the third layer of the framework, the local banknote features have been utilized. the fourth layer is a classification layer where different techniques are applied. Moreover, the developed system recognizes multiple banknotes. The implemented framework paves the way for various mobile applications are used effectively and commercially to assist visually impaired people.

Evaluation of Deep Learning Models for Detection of Indonesian Rupiah

“…Here, computer vision technology holds out the possibility of helping those who are blind due to its potential for advancement [9].…”