Search In this Thesis
   Search In this Thesis  
العنوان
An ATM identity authentication system based on biometrics /
المؤلف
Kassem, Mohamed Abdo Mohamed Ali.
هيئة الاعداد
باحث / محمد عبده محمد علي قاسم
مشرف / رشيد مختار العوضى
مشرف / نغم السيد مكي
مناقش / محمد حسن حجاج
الموضوع
Biometric identification. ATM. Biosensors.
تاريخ النشر
2015.
عدد الصفحات
132 p. :
اللغة
الإنجليزية
الدرجة
ماجستير
التخصص
Information Systems
تاريخ الإجازة
1/1/2015
مكان الإجازة
جامعة المنصورة - كلية الحاسبات والمعلومات - Department of Information Technology
الفهرس
Only 14 pages are availabe for public view

from 32

from 32

Abstract

Traditional biometric system based on a single biometric usually suffers from problems like ‘imposter’ attack or hacking unacceptable error rate and low performance, so the need of using multimodal biometric system is needed. The design of a multibiometric system is a challenging task due to heterogeneity of the biometric sources in terms of their type of information, the magnitude of information content, correlation among the different sources and conflicting performance requirements of the practical applications. In this thesis, four frameworks are developed at score and decision fusion in multibiometric systems. Database for the experimental is consist of two unimodal database one for the finger and the second for iris. Finger image is downloaded from the fingerprint competition 2004 (FVC 2004 DB1_B and DB3_B). Iris images is downloaded from the Chinese Academy of Science (CASIA Ver. 1.0). A multimodal fingerprint and iris identification and authentication system is presented. Features of both modalities are extracted by using minutia extraction for fingerprint and canny edge detection for iris. A comparison of two biometric modalities are developed. Score and decision level fusion for fingerprint and iris are developed to integrate with ATM for the hard secure to save the money from hackers. Minutia matcher is used for fingerprint and hamming distance for iris to compute the different between two irises template. The result obtained are 0%, 5%, 96.67% and 44 second for FAR, FRR, Accuracy and time respectively. It is a highly accepted with ATM. According to fingerprint computation time is less then iris. We use this property against iris. The finger needs only 15 seconds to get match but iris needs 29 seconds. A threshold is put on fingerprint to pass specific number of fingerprints with a big score than threshold and minutia alignment is add to do this part. Here the enrolled iris will compare with the irises related to fingerprints passes through threshold. By computing accuracy it was the same but differ in response time 0%, 5% and 96.67% for FAR, FRR, and Accuracy but it take 32 seconds only. It has been found that the third proposed multi modal system fused at the decision level are not acceptable because of its low accuracy of 75%. Another proposed system is developed here to get less time. The fourth proposed multi modal fusion system was at score level when user entering PIN. When user enter his recorded pin, ATM will operate directly on his saved templates linked with PIN to match these templates with the enrolled templates. The proposed multi modal system combines PIN with score level fusion has the best result. The results obtained from this system are 0%, 5%, 96.67% and 0.646 part of second for FAR, FRR, accuracy and run time. from the fourth proposed system the main motivation and contribution are achieved by accruing non-false accept rate with high accuracy rate and small time to identify and authenticate person to use the ATM.