We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
Recognition and Fitness Classification of Nigerian Currency for Automatic Teller Machines.
- Authors
Woods, Nancy Chinyere; Oladosu, Halimah Shadiat
- Abstract
Automatic teller machines (ATMs) are now in use almost everywhere, therefore naira notes need to be recognized and classified. Recognizing and classifying banknotes can be a tedious task when performed manually, because of the volume of notes to be evaluated. The problem not only consists of automatically recognizing banknotes by denominations, sides, and directions, but also in determining the fitness of those banknotes for an ATM. It is therefore important to correctly recognize banknotes as well as determine the notes' fitness for an ATM. In this work, a system was designed to recognize naira banknotes and also determine the notes fitness for use in an ATM. Naira notes were scanned and labelled as ATM fit or ATM unfit based on the Central Bank of Nigeria (CBN) banknote fitness guidelines. For every naira note image in the dataset, Speeded-Up Robust Features (SURF) was used to extract several keypoints of varying sizes and orientation and these were stored for reference. Given a test image, K-nearest neighbour (KNN) algorithm was used to match its features with pre-computed features to determine the note's denomination. The test note's fitness was then determined by calculating the image quality difference between the test note and a similar reference note using mean square error (MSE), peak signal-to-noise ratio (PSNR) and structural similarity index (SSIM). Our system achieved an accuracy of 100% for naira note denomination recognition. For ATM fitness sorting, an average accuracy of 97%, 75.2%, 81.2% was achieved by PSNR, SSIM, MSE respectively for fit naira notes. While for unfit naira notes, the average accuracy achieved was 93.8%, 88.7%, 93.2% by PSNR, SSIM, MSE respectively. The result obtained showed that the PSNR technique outperformed the SSIM and MSE and is therefore efficient and reliable for sorting of naira notes based on the CBN banknotes fitness guidelines.
- Subjects
NAIRA (Nigeria currency); AUTOMATED teller machines; CENTRAL Bank of Nigeria; SIGNAL-to-noise ratio; BANK notes; FEATURE extraction
- Publication
African Journal of Computing & ICT, 2020, Vol 13, Issue 1, p29
- ISSN
2006-1781
- Publication type
Article