CLAHE ve Sobel Filtreleriyle Geliştirilmiş İki Kanallı Hibrit Evrişimli Makine Öğrenmesi ile Parmak Damar İzi Sınıflandırması

Abstract

Dijital teknolojinin ilerlemesi, özellikle parmak damarı tespiti alanında biyometrik güvenlik sistemlerinin yükselişine sebep olmuştur. Literatürde parmak damarı sınıflandırması üzerine yapılan araştırmaların çoğunda yüksek doğruluk elde etmek ana amaç iken genelleme kapasitesi ve test dağılımı gibi konular genellikle göz ardı edilmektedir. Bu çalışmada, farklı test dağılımlarıyla iki farklı veri seti (MMCBNU_6000 ve FV-USM) K-Katlamalı yapı kullanılarak tarafsız örnekleme için test edilmiştir. Deney bölümünde, Kontrast Sınırlı Adaptif Histogram Eşitleme (KSAHE) ve Sobel filtreleme gibi iki farklı görüntü iyileştirme yöntemi veri setlerine uygulanmış ve özellik çıkarma için Evrişimli Sinir Ağları (ESA) kullanılmıştır. Ayrıca, sınıflandırma için makine öğrenimi algoritmaları uygulanmış ve Hibrit Evrişimli Makine Öğrenimi algoritması oluşturulmuştur. Bu yöntem, konvansiyonel öğrenme algoritmalarına kıyasla, iki farklı kanal ile beslenen model, klasik makine öğrenmesi sınıflandırıcıları ile ESA modelini birleştirmektedir. Bu doğrultuda çalışmada üç görev belirlenmiştir: ilk iki görevde her bir veri kümesi için çeşitli makine öğrenimi algoritmalarının uygulanması odaklanmışken, üçüncü görev veri kümelerinin birleştirilmesi ve Yığma Topluluk Sınıflandırıcısı (YTS) kullanımını içermiştir. Modellerin değerlendirilmesinde doğruluk ve F1-skoru metrikleri kullanılmıştır. Sonuçlar, en yüksek doğruluk skorunun %98.94 ile üçüncü deneyle elde edildiğini göstermektedir. Ayrıca test verisi sayısının artmasının (%20 Test ve %50 Test arasındaki fark) modelin doğruluk metriğinde önceki çalışmalara kıyasla minimal bir düşürme etkisine sahip olduğu gözlemlenmektedir.

Keywords

• Evrişimli Sinir Ağları
• Makine Öğrenmesi
• Yığma Topluluk Öğrenmesi
• K- Katlamalı Çapraz Doğrulama
• Parmak Damar İzi

Enhancing Finger Vein Classification through CLAHE and Sobel Filtering with Two Channel Hybrid Convolutional Machine Learning Algorithm

Abstract

Advancements in digital technology have driven the rise of biometric security systems, notably in the field of finger vein detection. In most of the research on finger vein classification in the literature, achieving high accuracy is the main aim, while aspects such as generalization capacity and test distribution are mostly overlooked. In this study, two different datasets (MMCBNU_6000 and FV-USM) were tested with different test distributions, using a K-Fold structure for unbiased sampling in classification. In experiment part, two distinct image enhancement methods, namely Contrast Limited Adaptive Histogram Equalization (CLAHE) and Sobel filtering, were utilized on the datasets, and Convolutional Neural Networks (CNN) were used for feature extraction. Furthermore, machine learning algorithms were applied for classification, forming a Hybrid Convolutional Machine Learning algorithm. In this method, the model, which is fed through two different channels compared to conventional learning algorithms, combines classical machine learning classifiers with the CNN model. In the scope of this study, three tasks were outlined. The first two focused on implementing various machine learning algorithms for each dataset, while the third involved merging datasets and employing the Stacking Ensemble Classifier (SEC). For evaluating the models, accuracy and F1-score metrics were used. The results indicate that the highest accuracy rate was achieved in the third experiment, with a score of 98.94%. Additionally, it is also observed that increasing the amount of test data (the difference between 20% Test and 50% Test) has a minimal effect in reducing the model's accuracy metric compared to previous studies.

Keywords

• Convolutional Neural Networks
• Machine Learning
• Stacking Ensemble Learning
• K-Fold Cross Validation
• Finger Vein

References

1. [1] Shaheed, K., Mao, A., Qureshi, I., Kumar, M., Hussain, S., Zhang, X. 2022. Recent advancements in finger vein recognition technology: methodology, challenges and opportunities, Inf. Fusion, Vol. 79, pp. 84–109.
2. [2] Lian, F.-Z., Huang, J.-D., Liu, J.-X., Chen, G., Zhao, J.-H., Kang, W.-X. 2023. FedFV: A personalized federated learning framework for finger vein authentication, Mach. Intell. Res., Vol. 20, No. 5, pp. 683–696.
3. [3] Zhang, L., Li, W., Ning, X., Sun, L., Dong, X. 2021. A local descriptor with physiological characteristic for finger vein recognition, in: 2020 25th International Conference on Pattern Recognition (ICPR), pp. 4873–4878. DOI: 10.1109/ICPR48806.2021.9412203.
4. [4] Zhang, L., et al. 2022. A joint Bayesian framework based on partial least squares discriminant analysis for finger vein recognition, IEEE Sens. J., Vol. 22, No. 1, pp. 785–794. DOI: 10.1109/JSEN.2021.3130951.
5. [5] Lv, W., Ma, H., Li, Y. 2023. A finger vein authentication system based on pyramid histograms and binary pattern of phase congruency, Infrared Phys. Technol., Vol. 132, p. 104728. DOI: 10.1016/j.infrared.2023.104728.
6. [6] Yang, J., Shi, Y., Yang, J., Jiang, L. 2009. A novel finger-vein recognition method with feature combination, in: 2009 16th IEEE International Conference on Image Processing (ICIP), pp. 2709–2712.
7. [7] Boucherit, I., Zmirli, M.O., Hentabli, H., Rosdi, B.A. 2022. Finger vein identification using deeply-fused convolutional neural network, J. King Saud Univ. Comput. Inf. Sci., Vol. 34, No. 3, pp. 646–656. DOI: 10.1016/j.jksuci.2020.04.002.
8. [8] Pizer, S.M. 1990. Contrast-limited adaptive histogram equalization: speed and effectiveness, in: Proceedings of the First Conference on Visualization in Biomedical Computing, Atlanta, Georgia, p. 2.

9. [9] Zhang, L., Wang, X., Dong, X., Sun, L., Cai, W., Ning, X. 2021. Finger vein image enhancement based on guided tri-Gaussian filters, ASP Trans. Pattern Recognit. Intell. Syst., Vol. 1, No. 1, pp. 17–23.
10. [10] Kilic, U., Karabey Aksakalli, I., Tumuklu Ozyer, G., Aksakalli, T., Ozyer, B., Adanur, S. 2023. Exploring the effect of image enhancement techniques with deep neural networks on direct urinary system (DUSX) images for automated kidney stone detection, Int. J. Intell. Syst., Vol. 2023, No. 1, p. 3801485.
11. [11] Wang, Y., Lu, H., Qin, X., Guo, J. 2023. Residual Gabor convolutional network and FV-Mix exponential level data augmentation strategy for finger vein recognition, Expert Syst. Appl., Vol. 223, p. 119874.
12. [12] Hou, B., Yan, R. 2021. ArcVein-Arccosine center loss for finger vein verification, IEEE Trans. Instrum. Meas., Vol. 70, pp. 1–11. DOI: 10.1109/TIM.2021.3062164.
13. [13] Zhang, Y., Liu, Z. 2020. Research on finger vein recognition based on sub-convolutional neural network, in: 2020 International Conference on Computer Network, Electronic and Automation (ICCNEA), pp. 211–216. DOI: 10.1109/ICCNEA50255.2020.00051.
14. [14] Kuzu, R.S., Maiorana, E., Campisi, P. 2020. Vein-based biometric verification using transfer learning, in: 2020 43rd International Conference on Telecommunications and Signal Processing (TSP), pp. 403–409. DOI: 10.1109/TSP49548.2020.9163491.
15. [15] Tao, Z., Zhou, X., Xu, Z., Lin, S., Hu, Y., Wei, T. 2021. Finger-vein recognition using bidirectional feature extraction and transfer learning, Math. Probl. Eng., Vol. 2021, p. 6664809. DOI: 10.1155/2021/6664809.
16. [16] Vaswani, A., et al. 2017. Attention is all you need, Adv. Neural Inf. Process. Syst., Vol. 30.
17. [17] Dosovitskiy, A., et al. 2020. An image is worth 16x16 words: transformers for image recognition at scale, ArXiv Prepr. ArXiv201011929.
18. [18] Zhao, P., et al. 2024. VPCFormer: A transformer-based multi-view finger vein recognition model and a new benchmark, Pattern Recognit., Vol. 148, p. 110170. DOI: 10.1016/j.patcog.2023.110170.
19. [19] Li, X., Zhang, B.-B. 2023. FV-ViT: Vision transformer for finger vein recognition, IEEE Access.
20. [20] Garcia-Martin, R., Sanchez-Reillo, R. 2023. Vision transformers for vein biometric recognition, IEEE Access, Vol. 11, pp. 22060–22080.
21. [21] An, Z., Ren, X., Tao, Z. 2024. FV-DMHN: Dual multi-head network for finger vein recognition, IEEE Access.
22. [22] Lu, Y., Xie, S.J., Yoon, S., Wang, Z., Park, D.S. 2013. An available database for the research of finger vein recognition, in: 2013 6th International Congress on Image and Signal Processing (CISP), pp. 410–415. DOI: 10.1109/CISP.2013.6744030.
23. [23] Lu, Y., Xie, S.J., Yoon, S., Yang, J., Park, D.S. 2013. Robust finger vein ROI localization based on flexible segmentation, Sensors, Vol. 13, No. 11, pp. 14339–14366. DOI: 10.3390/s131114339.
24. [24] Mohd Asaari, M.S., Suandi, S.A., Rosdi, B.A. 2014. Fusion of band limited phase only correlation and width centroid contour distance for finger based biometrics, Expert Syst. Appl., Vol. 41, No. 7, pp. 3367–3382. DOI: 10.1016/j.eswa.2013.11.033.
25. [25] Cortes, C., Vapnik, V. 1995. Support-vector networks, Mach. Learn., Vol. 20, No. 3, pp. 273–297. DOI: 10.1007/BF00994018.
26. [26] Wolpert, D.H. 1992. Stacked generalization, Neural Netw., Vol. 5, No. 2, pp. 241–259. DOI: 10.1016/S0893-6080(05)80023-1.
27. [27] Pedregosa, F. 2011. Scikit-learn: Machine learning in Python, J. Mach. Learn. Res., Vol. 12, p. 2825.