Evaluation of the Proposed Hand Vein Authentication System Using Machine Learning - Article 2

Aug 5, 2026

Upload by Skyler

This paper evaluates a biometric authentication approach that leverages deeper exploration of hand vein patterns to increase reliability for lifelong identification and verification. The proposed workflow uses a three-phase pipeline: data processing, feature extraction, and feature evaluation. Advanced filtering on a dataset of 420 hand vein images applies Box filters, HFEF, and CLAHE, followed by discrete cosine transformation and image fusion. PCA Net extracts distinctive vein features, then classification is performed using SVM, Logistic Regression, and Naive Bayes, achieving the best accuracy with Logistic Regression (99.7%) and SVM (99.6%).

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages14

Reading Time35 min

AccessFree to download

Abstract

Introduction

Biometric system overview

Hand vein as a biometric trait

Related Work

Palm vein recognition overview

Near-infrared vein identification

Academic Science Journal Evaluation of the Proposed Hand Vein Authentication System Using Machine Learning Rajaa Ahmed Ali 1*and Ziyad Tariq Mustafa Al-Ta’i 2 1Department of Computer Science, The Institute of informatics for post-graduation, University of Information Technology and Communication 2Department of Computer Science, College of Science, University of Diyala *[Phd202120689@iips.edu.iq](Phd202120689@iips.edu.iq) This article is open-access under the CC BY 4.0 license)[http://creativecommons.org/licenses/by/4.0](http://creativecommons.org/licenses/by/4.0)( Received: 6 December 2024 Accepted: 7 January 2025 Published: 28 April 2025 DOI: [https://dx.doi.org/10.24237/ASJ.03.02.960B](https://dx.doi.org/10.24237/ASJ.03.02.960B) Abstract This paper discusses the use of a biometric system for secure authentication based on deeper exploration in hand vein patterns to derive more reliability as a good biometric trait. The proposed system identifies internal hand vein patterns, and also identifies individuals correctly throughout their lifetimes. The proposed system follows a three-phase approach comprising data processing, feature extraction, and feature evaluation. For the advanced filtering techniques ina dataset of 420 hand vein images, the approach goes to Box filters, HFEF, and CLAHE, then to discrete cosine transformation and image fusion. The features are extracted through PCA Net for acquiring the most distinctive attributes of hand veins. The different machine learning algorithms used in this evaluation for classification of the extracted features include SVM, Logistic Regression, and Naive Bayes. Results indicate the highest accuracy for the Logistic Regression algorithm (99.7%) and the SVM algorithm of (99.6%). However, the Random Forest algorithm has an accuracy of (98%), while the Naive Bayes algorithm shows a poorer accuracy of (91%) . Keywords: Biometric authentication, hand vein patterns, machine learning, feature extraction. Volume: 3, Issue: 2, April 2025 P-ISSN: 2958-4612 E-ISSN: 2959-5568 Academic Science Journal Introduction Biometric systems, on the other hand, define or authenticate individuals by using their biological attributes or their behavioral characteristics measured against biometric designs in the database of similar attributes. Hence, this measurement can be classified as an identification or verification system and can be defined as a computerized method of identifying one's identity or verifying it through specific biological or behavioral characteristics [1] . These are the two unique kinds of biometric characteristics that do not vary over time: physiological and behavioral. The physiological features include the fingerprint, DNA, iris, hand, face, etc., while behavior includes voice, signature, and keystroke, etc., as shown in Figure 1. [2] . Figure 1 :Biometric systems general block diagram [2] . The fact is that the palm vein pattern be used as a biometric trait for authentication as this pattern has spatial geometry of variable dimensions for each user which cannot be amended [3] . The interior part of the hand behind the thumb and index finger is called the palm. The image of the correct hand held on the flat, glass surface of a scanner is used to extract the finger feature, Volume: 3, Issue: 2, April 2025 P-ISSN: 2958-4612 E-ISSN: 2959-5568 Academic Science Journal hand geometry, and palm features. In order to ensure sufficient scanning quality, the user will be required to place their hand on the scanner [4] . Several physiological features can be extracted from the hand and used as biometrics. Of these, the most popular humanoid biometric method is using hand veins, which has characteristic advantages. Since their source lies internal and inaccessible to the body, these are comparatively much more secure and difficult to forge. Furthermore, the capturing of vein patterns is possible only when a subject is alive, thus excluding the

FAQs

What is the main goal of the proposed hand vein authentication system?

The system aims to provide secure authentication by extracting and evaluating internal hand vein patterns to achieve reliable identification and verification over a lifetime.

What are the three phases of the proposed workflow?

The pipeline includes data processing, feature extraction, and feature evaluation to prepare images, learn distinctive vein features, and classify identities.

Which machine learning algorithms were used for classification, and what were the results?

The evaluation uses SVM, Logistic Regression, and Naive Bayes for classification. Logistic Regression achieved the highest accuracy (99.7%), followed closely by SVM (99.6%), while Random Forest (98%) and Naive Bayes (91%) were lower.

Aug 5, 2026

Upload by Skyler