Comparison of Gabor Filter Parameter Characteristics for Dorsal Hand Vein Authentication Using Artificial Neural Networks

Abstract

The importance of digital security in today's technological era requires various innovations in creating a reliable security system for humans. Biometrics is an authentication method and the most effective system for performing personal recognition because biometrics have unique characteristics. Dorsal hand vein become biometrics for the individual recognition process in this study using feature extraction of gabor filters and neural network backpropagation to classify recognition into five classes of human individuals, which are expected to be able to provide a higher accuracy value when compared to research on the introduction of dorsal hand vein. This classification process has several stages, namely input image, image pre-processing, segmentation, feature extraction, and image classification. The test results show that the percentage of success based on the five test scenarios has an average value of 75%. In this study, the results of the greatest test accuracy in the fourth scenario were 91%.