Enhancing Finger Vein Recognition with Gaussian CLAHE Preprocessing: A Comprehensive Analysis on FV-USM Database

Finger vein recognition has emerged as a promising biometric authentication technique due to its high security and contactless nature. This paper presents a new approach to enhance finger vein recognition performance using ResNet18 architecture with Gaussian Blur and Contrast Limited Adaptive Histogram Equalization (CLAHE) preprocessing on the Finger Vein USM (FV-USM) Database. We conducted extensive experiments comparing baseline ResNet18 models with Gaussian CLAHE-enhanced counterparts across four individual finger configurations and their combinations. The results demonstrate remarkable improvements with Gaussian CLAHE preprocessing, with accuracy increasing from <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$36.7-48.6 \%$</tex> to <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$86.0-90.1 \%$</tex> for individual fingers and from 59.9 % to 87.1 % for the combined approach. Our analysis reveals that fingers 3 and 4 yield the highest recognition rates (89.6 % and 90.1 %, respectively) after enhancement, even surpassing the combined finger approach. Visual examination confirms that Gaussian CLAHE preprocessing substantially enhances vascular pattern visibility, directly contributing to improved feature extraction. This study provides valuable insights into finger vein recognition enhancement through preprocessing techniques and offers a strong foundation for developing more robust biometric authentication systems.