AAU-ETD AAU-ETD
 

Addis Ababa University Libraries Electronic Thesis and Dissertations: AAU-ETD! >
Faculty of Technology >
Thesis - Computer Engineering  >

Please use this identifier to cite or link to this item: http://hdl.handle.net/123456789/2971

Title: MODELLING BANKNOTE AND COIN RECOGNITION SYSTEMS USING NEURAL NETWORK A Case Study for Ethiopia
Authors: Tewodros, Beyene Mesfin
Advisors: Dr.Eng. Getachew Alemu
Keywords: banknote, coin, PCA, LDA, Gabor
moment invariant.
MLP
UV
Copyright: Feb-2011
Date Added: 11-May-2012
Publisher: AAU
Abstract: n this thesis, we dealt on recognition of banknotes and coins of Ethiopic ones. There are five types of Ethiopian banknotes; these are 1birr, 5birr, 10birr, 50birr, and 100birr and four insertion directions are considered for each note, namely front up, front down, backup and back down. Hence; we have a total of 20 images to be recognized. For the case of coins, there are 4 types of coins available and for the classification task we considered only head view of each coin and insertion direction for coins is made to be at any degree of rotation of a coin. The main concern in recognition of coins for this thesis is on analyzing the head images obtained through image acquisition process. However, the rest of parameters such as weight and thickness values of coins are assumed to be measured accurately. In addition to the classification tasks gone through, the check up of counterfeit notes has also been one part of this thesis. In the image taking process, there are two possible image acquisition methods, one is by using normal digital scanner, the other is using Ultraviolet (UV) scanner, and the second method has been concluded to be more appropriate for the counterfeit identification task from some previous work. Putting this result into consideration, we also used the second method; however, instead of taking the whole UV images for analysis, we considered certain counterfeit measure regions of the UV images and this has a considerable advantage on minimization of computation time and complexity. In the recognition task, different steps are gone through and these are image acquisition, preprocessing, feature extraction and finally classification of birr notes using multilayer perceptron(MLP). The preprocessing stage consists of various image contrast enhancement techniques and noise removal methods. In the feature extraction stage, techniques such as Principal Components Analysis (PCA), Linear Discriminant Analysis (LDA) and Gabor filter are used to lower dimension of images so as to ease the computation complexity as well lower computation time. During the course of analysis, system with LDA is found to have better recognition rate of 97.5% for different noise levels considered. The Gabor one is optimal while dealing with tilted and slanted images of birr notes because of its rotation invariant nature. For the case of coins, Moment Invariant method is used for feature extraction and rotation invariance is achieved quite perfectly.
URI: http://hdl.handle.net/123456789/2971
Appears in:Thesis - Computer Engineering

Files in This Item:

File Description SizeFormat
320994595926431753799956781495822251494.2 MBAdobe PDFView/Open

Items in the AAUL Digital Library are protected by copyright, with all rights reserved, unless otherwise indicated.

 

  Last updated: May 2010. Copyright © Addis Ababa University Libraries - Feedback