Paper Key : IRJ************051
Author: N. Nibishaa
Date Published: 02 Oct 2023
Abstract
Numerous algorithms for finger vein feature extraction demonstrate commendable performance by emphasizing texture characteristics, yet some overlook the intensity distribution of the finger tissue, treating it as background noise in certain instances. Use this kind of noise as a novel soft biometric feature in this project to achieve better output in finger vein recognition. To begin, a thorough examination of the finger vein imaging theory and image characteristics is presented, showcasing the potential extraction of the intensity distribution generated by the finger tissue background for identification as a soft biometric feature. Subsequently, this study introduces two algorithms for extracting the background layer of finger vein patterns and three algorithms for extracting soft biometric traits from the intensity distribution. In the classification stage developed a system with implementation of convolution neural network specifically RESNET18 for the training image dataset and image retrieving process is done. Purpose of introducing deep learning in developing finger vein identification system is to get accurate more performance and speedy results. These are computed on the basis Euclidean distance between features obtained from test image and features of trained images, the model designed has good robustness in illumination and rotation.
DOI LINK : 10.56726/IRJMETS45027 https://www.doi.org/10.56726/IRJMETS45027