Finger Vein Based CNN Algorithms for Human Recognition

Abstract

Due to factors like data confidentiality insurance and higher accuracy, finger vein-based systems are obtaining extra attention in current biometric security systems. The majority of previous research relied on palm veins, fingerprints, and other biometrics. Due to the location of finger veins behind the skin which are both more secure than fingerprint systems and uniquely different for each person, they cannot be used for falsification. This paper discusses the use of CNN algorithms in finger vein recognition systems. In our work, we start with the preprocessing the images of the finger veins must first be preprocessed. After that two models were chosen specifically for the feature extraction and classification stage, which recognizes individual identification. The detailed collection of tests demonstrates that the accuracy possible with the suggested method can achieve a 99% correct identification rate for SDUMLA HMT data.