Practical implementation of the Siamese neural network in biometric palmar digital veins authentication

Authors: Desyatov A.G.
Published in issue: #5(82)/2023
DOI: 10.18698/2541-8009-2023-5-892
Category: Informatics, Computer Engineering and Control | Chapter: Methods and Systems of Information Protection, Information Security
Keywords: biometrics, authentication, palmar digital veins, deep learning, Siamese neural network, Conv2D, MaxPooling2D, Dropout, Dense, dataset
Published: 19.05.2023

The paper presents theoretical foundations of the Siamese neural network and considers its construction to implement the biometric palmar digital veins authentication in detail. The SDUMLA-HMT palmar digital veins open dataset was analyzed, which assisted in forming the samples of learning and test data. The paper describes practical implementation of learning a Siamese neural network based on the generated samples. For various thresholds, probabilities of type I and type II errors, which were the key metrics in biometric authentication, were calculated. Results of introducing the developed deep learning model were analyzed and are presented in the form of graphs.

References

[1] Introducing Hitachi’s newest addition to the digital security portfolio, VeinID Five. URL: https://digitalsecurity.hitachi.eu/products/veinid-five/ (accessed March 10, 2023).

[2] (Group 3) Siamese Face Recognition. URL: https://openpower.ucc.in.tum.de/group-3-siamese-face-recognition/ (accessed March 13, 2023).

[3] Li B., Yan J., Wu W., Zhu Z., Hu X. High Performance Visual Tracking with Siamese Region Proposal Network. IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Salt Lake City, UT, USA, IEEE, 2018. http://dx.doi.org/10.1109/CVPR.2018.00935

[4] SDUMLA-HMT: A Multimodal Biometric Database. URL: https://link.springer.com/chapter/10.1007/978-3-642-25449-9_33 (accessed March 17, 2023).

[5] High Performance Visual Tracking with Siamese Region Proposal Network. URL: http://openaccess.thecvf.com/content_cvpr_2018/papers/Li_High_Performance_Visual_CVPR_2018_paper.pdf (accessed March 15, 2023).

[6] Building image pairs for Siamese networks with Python. URL: https://pyimagesearch.com/2020/11/23/building-image-pairs-for-siamese-networks-with-python/ (accessed March 23, 2023).

[7] Keras: biblioteka glubokogo obuchenie na Python [Keras: Python deep learning library]. URL: https://ru-keras.com/home/ (accessed March 22, 2023).

[8] Podrobnoe ob"yasnenie siamskoy seti i sravnitel’noy poteri [Detailed explanation of Siamese network and comparative loss]. URL: https://russianblogs.com/article/74731636351/ (accessed March 23, 2023).

[9] Training & evaluation with the built-in methods. URL: https://keras.io/guides/training_with_built_in_methods/ (accessed March 23, 2023).

[10] Glushchenko N.A., Konnova N.S. Neural network approach to verifying manual signature. Politekhnicheskiy molodezhnyy zhurnal, 2018, no. 5. http://dx.doi.org/10.18698/2541-8009-2018-5-313

[11] Ling Jin. Using Deep Learning for finger-vein based biometric authentication. URL: https://towardsdatascience.com/using-deep-learning-for-finger-vein-based-biometric-authentication-3f6601635821 (accessed March 23, 2023).