Fime study X-rays impact on fingerprint biometric systems from environmental conditions
Joël Di Manno, Authentication and Biometrics Laboratory Service Line manager and Abdarahmane Wone, Biometrics and Artificial Intelligence researcher at Fime, have published a scientific paper in which they examine the bearing which environmental conditions have on the performance of biometric authentication systems.
The paper contains findings of a study conducted by Fime to better understand how such environmental conditions affect the performance of those systems, and to propose how such challenges can be overcome.
Explaining the details of their paper in an interview published on the blog section of the company’s website, the Fime officials posited that while using a person’s biometric characteristics to authenticate them provides high level security and trust, some environmental conditions can drastically alter the performance of some of these systems, leading to either non-genuine users getting verified or genuine ones not being able to.
“For example, if someone is using a facial recognition solution, changes in lighting or the background can influence its performance. Similarly, fingerprint systems can be affected when environmental temperature and humidity change because the texture of fingerprints alter accordingly… These environmental changes impact the performance, security, user experience and trust of biometric systems,” says Abdarahmane.
On the study’s findings, Abdarahmane explains: “We tested the performance of three different third-party fingerprint authentication matchers in different climatic conditions. The aim was to see how accurate the algorithms were at matching the fingerprint samples taken during enrolment. The performance of the biometric systems was evaluated in six different conditions made up of a combination of two different temperatures and three different humidity environments. The different humidity and temperature environments were created using climatic chambers.”
The study was based on analysis of more than a thousand fingerprint images collected from 17 volunteers.
According to the official, the findings show algorithms of the fingerprint biometric system “performed better when the temperature was less humid” and that “the three algorithms were all impacted differently by temperature and humidity changes, demonstrating that the impact of environmental factors is not consistent across biometric solutions.”
Also important to note, Abdarahmane, said is the fact that all the algorithms performed better when the environmental conditions were the same as those during enrolment of the fingerprint samples, thus highlighting the importance of a comprehensive enrolment guide for vendors and users, to decrease the impact of environmental conditions as much as possible.
Also in the interview, Di Manno stated that thanks to the findings of the study, Fime has now been able to develop a process and identify factors for the proper evaluation of environmental impact on biometric authentication systems.
“The results of this research demonstrate that environmental conditions can have differing degrees of impact on biometric authentication systems. Therefore, testing the performance of biometric solutions in different environments, including different conditions between enrolment and verification, could prevent real-life issues. Certification schemes could introduce this aspect into their evaluation programs to ensure security in various conditions and decrease variance between different biometric solutions,” says Di Manno.
The paper is recommended for use by biometric solution vendors in their quality assurance processes.
accuracy | authentication | biometric enrollment | biometric identification | biometric testing | biometrics | biometrics research | Fime | fingerprint recognition