Image based ringgit banknote recognition for visually impaired

• Main Authors: Jasmin Sufri, N. A., Rahmad, N. A., As'ari, M. A., Zakaria, N. A., Jamaludin, M. N., Ismail, L. H., Mahmood, N. H.
• Format: Article
• Published: Universiti Teknikal Malaysia Melaka 2017
• Subjects: TK Electrical engineering. Electronics Nuclear engineering
• Online Access: http://eprints.utm.my/id/eprint/76613/; https://www.scopus.com/inward/record.uri?eid=2-s2.0-85041929180&partnerID=40&md5=2fbc2a17f7ceab9331802aa56b5d6687
• Institution: Universiti Teknologi Malaysia

Visually impaired people face a number of difficulties in order to interact with the environment because most of the information encoded is visual. Visual impaired people faced a problem in identifying and recognizing the different currency. There are many devices available in the market but not acceptable to detect Malaysian ringgit banknote and very pricey. Many studies and investigation have been done in introducing automated bank note recognition system and can be separated into vision based system or sensor based system. The objective of this project was to develop an automated system or algorithm that can recognize and classify different Ringgit Banknote for visually impaired person based on banknote image. In this project, the features extraction of the RGB values in six different classes of banknotes (RM1, RM5, RM10, RM20, RM 50, and RM100) was done by using Matlab software. Three features called RB, RG and GB extracted from the RGB values were used for the classification algorithms such as k-Nearest Neighbors (k-NN) and Decision Tree Classifier (DTC) for recognizing each classes of banknote. Ten-fold cross validation was used to select the optimized k-NN and DTC, which was based on the smallest cross validation loss. After that, the performance of optimize k-NN and DTC model was presented in confusion matrix. Result shows that the proposed k-NN and DTC model managed to achieve 99.7% accuracy with the RM50 class causing major reduction in performance. In conclusion, an image based automated system that can recognize the Malaysian banknote using k-NN and DTC classifier has been successfully developed.