Affiliation:
1. University of Notre Dame
Abstract
Iris recognition has been revolutionising biometric-based human identification for almost three decades, due to the unprecedented information richness of iris patterns translating into low false matches – which has now been observed across a population exceeding 1 billion users worldwide. We are also able to effectively detect those presentation attack types known to our algorithms during the design phase. But the question remains: are we prepared to detect those unknown specimens, strikingly similar to images of authentic irises, that can be generated by modern artificial intelligence tools or – more gruesomely – captured from decedents?
Subject
Electrical and Electronic Engineering,General Computer Science
Cited by
4 articles.
订阅此论文施引文献
订阅此论文施引文献,注册后可以免费订阅5篇论文的施引文献,订阅后可以查看论文全部施引文献