TY - GEN
T1 - Fingerprint Identification from Digital Images Using Deep Learning
AU - Khaled, M. Moneb
AU - Sayadi, Aghyad A.L.
AU - Alsmirat, Mohammad
AU - Al-Ayyoub, Mahmoud
N1 - Publisher Copyright:
© 2023 IEEE.
PY - 2023
Y1 - 2023
N2 - Authentication methods, particularly those based on biometrics, are becoming increasingly popular due to their superior security, cost-effectiveness, and user-friendliness compared to conventional methods. The spread of contagious illness has led to the urgent need for contactless fingerprint verification in various sectors. However, challenges arise in fingerprint categorization in touchless systems due to factors like image clarity, background interference, besides external conditions. This research introduces an initial work on a deep learning system for touchless fingerprint identification, utilizing a comprehensive dataset of 2,143 images from 175 participants. The task is presented as a classification problem. Our proposed solution combines preprocessing strategies with deep and transfer learning models, incorporating various preprocessing methods to boost the classification efficacy in touchless fingerprint identification. Based on our tests, the InceptionResnet model emerged as the top performer, registering an accuracy rate of 89%.
AB - Authentication methods, particularly those based on biometrics, are becoming increasingly popular due to their superior security, cost-effectiveness, and user-friendliness compared to conventional methods. The spread of contagious illness has led to the urgent need for contactless fingerprint verification in various sectors. However, challenges arise in fingerprint categorization in touchless systems due to factors like image clarity, background interference, besides external conditions. This research introduces an initial work on a deep learning system for touchless fingerprint identification, utilizing a comprehensive dataset of 2,143 images from 175 participants. The task is presented as a classification problem. Our proposed solution combines preprocessing strategies with deep and transfer learning models, incorporating various preprocessing methods to boost the classification efficacy in touchless fingerprint identification. Based on our tests, the InceptionResnet model emerged as the top performer, registering an accuracy rate of 89%.
KW - Biometric authentication
KW - Contactless fingerprint recognition
KW - Deep learning
KW - Touchless fingerprint classification
KW - Transfer learning
UR - https://www.scopus.com/pages/publications/85182732654
U2 - 10.1109/ICSC60084.2023.10349980
DO - 10.1109/ICSC60084.2023.10349980
M3 - Conference contribution
AN - SCOPUS:85182732654
T3 - 2023 3rd Intelligent Cybersecurity Conference, ICSC 2023
SP - 26
EP - 31
BT - 2023 3rd Intelligent Cybersecurity Conference, ICSC 2023
A2 - Jararweh, Yaser
A2 - Alsmirat, Mohammad
A2 - Aloqaily, Moayad
A2 - Alsmadi, Izzat
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 3rd Intelligent Cybersecurity Conference, ICSC 2023
Y2 - 23 October 2023 through 25 October 2023
ER -