Short Term Template Aging Effects on Biometric Dynamic Handwriting Authentication Performance

摘要

In biometrics the variance between data acquired from the same user and same trait is not only based on different sensors or user's form of the day, but it also depends on an aging factor. Over time the biological characteristics of a human body changes. This leads to physical and mental alternations, which may have significant influence on the biometric authentication process. In order to parameterize a biometric system, the study of the degree of aging's influence is an important step. In this paper we provide an experimental evaluation on the influence of changes of handwriting biometrics by acquiring data from writers in three sessions with a time difference of one month each. The aim is to analyze the potential impact of aging processes on different written content within a biometric handwriting system in terms of authentication performance. In the worst case, the equal error rate determined on verification data acquired two month after the reference data (EER = 0.162) is four times higher than the equal error rate calculated based on reference and verification data from the first session (EER = 0.041).