Finger Vein Verification Techniques using Convolutional Neural Networks
Lekshmi. J1, Gisha. G. S2
1Lekshmi.J*, Dept. of Computer Science & Engineering, LBSITW, Poojappura, TVPM, India.
2Gisha.G.S, Dept. of Computer Science & Engineering, LBSITW, Poojappura, TVPM, India.
Manuscript received on May 01, 2020. | Revised Manuscript received on May 07, 2020. | Manuscript published on May 30, 2020. | PP: 1514-1519 | Volume-9 Issue-1, May 2020. | Retrieval Number: A2469059120/2020©BEIESP | DOI: 10.35940/ijrte.A2469.059120
Open Access | Ethics and Policies | Cite | Mendeley
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: Finger vein beneath our skin is one of the unique features for identifying an individual. Because of its uniqueness and security the finger vein recognition is considered as a powerful biometric identifier for user authentication. Several techniques have been evolved for finger vein recognition from its early stage of development, but majority approaches were based on hand crafted features which had limitations on quality of the image, positioning of the finger etc. The emergence of neural networks led to the development of various Convolutional Neural Networks (CNN) based approaches for identity verification. This paper surveys various finger vein verification techniques using CNN and determines the factors that will affect the final result. Publicly available finger vein datasets as well as user designed ones, which are of different qualities, are used for the experimental analysis of these techniques. Though CNN is used in all the cases each one differs in the number of layers used, weight updating methods, results obtained etc. It is found that higher recognition accuracy and lower equal error rate (EER) makes the finger vein verification system an effective one. This field has emerged wide popularity recently and is used in different applications where security is of prime importance.
Keywords: Biometric identifier, Convolutional neural networks, Finger vein verification.
Scope of the Article: Convolutional Neural Networks