Multiple Instances Fingerprint Image Data Acquisition

Abstract

Using public fingerprint databases to validate the multi-instance fusion approach of a Multi-instance Biometric Authentication System (MBAS) for accurate authentication is a means of overcoming some of the limitations of a Unimodal Biometric System (UBS). Nevertheless, a significant portion of the web databases used for MBAS (Machine Learning-Based Automated Systems) were obtained under controlled conditions, with the photos being curated to align with a certain algorithmic objective. The performance of biometric systems exhibits variability when the datasets used in the algorithms undergo modifications as a result of discrepancies in the settings under which the photos were obtained. In this study, a database containing multiple fingerprint instances was developed locally in an uncontrolled environment. The aim was to create a database containing numerous examples of fingerprints for the same person from one or multiple data collection sessions. The database was produced locally by acquiring six samples of ten fingerprint instances of 150 subjects in an uncontrolled environment using a Futronic fingerprint scanner. The created database and results will be helpful to all researchers in the biometrics field as a tool for improving multi-instance fusion methods and enabling an unbiased evaluation of algorithms.