Longitudinal Evaluation of Child Fingerprint Recognition: Time Interval Up to Eight Years

Abstract

Child fingerprint recognition systems face unique challenges due to the rapid physiological changes associated with growth, making long-term recognition difficult. This study is focused on a comprehensive evaluation of fingerprint recognition performance over time, utilizing a longitudinal dataset comprising approximately 40K fingerprint images collected every six months over 8 years from 267 children aged 4 to 18 years. With the dataset, this study focused on the impact of aging, i.e., the variations in accuracy due to the increasing time gap between reference and probe samples. Additionally, the study assessed the effectiveness of customized Preciozzi’s scaling and VeriFinger’s fingerprint scaling methods, designed to adjust a fingerprint to compensate for changes related to the growth of a fingerprint. Despite significant aging effects in the 4–8-year age group, our findings reveal that a 4-year-old child remains recognizable for up to 48 months with a TAR of 97.4% at 0.1% FAR and up to 60 months with a TAR of 90.9% at 0.1% FAR without any scaling due to growth. Scaling techniques effectively mitigated decreases in performance, maintaining a TAR of 99.3% at 0.1% FAR for up to 90 months. Additionally, the study highlights that scaling is not strictly essential for children over 8 years of age up to 96 months but remains beneficial for achieving near-perfect accuracy (99–100%) over extended periods, especially for the 9–11-year age group. Linear Mixed-Effects Models revealed key predictors of comparison score variability, including time gap, enrollment age, and image quality, while accounting for their interactions and contributions to recognition performance. This study offers valuable insights into the longitudinal performance of child fingerprint recognition systems, bridging gaps in biometric research and paving the way for policy development and advanced technical solutions to enhance child safety and identification worldwide.