Seeded Ising Model and Distributed Biometric Template Storage and Matching

  • Choi, Hyeong In
  • Lee, Sungjin
  • Moon, Hwan Pyo
  • Wee, Nam-Sook
  • Kim, Daehoon
  • 외 1명
Citations

WEB OF SCIENCE

3
Citations

SCOPUS

3

초록

It is known that a variant of Ising model, called Seeded Ising Model, can be used to recover the information content of a biometric template from a fraction of information therein. The method consists in reconstructing the whole template, which is called the intruder template in this paper, using only a small portion of the given template, a partial template. This reconstruction method may pose a security threat to the integrity of a biometric identity management system. In this paper, based on the Seeded Ising Model, we present a systematic analysis of the possible security breach and its probability of accepting the intruder templates as genuine. Detailed statistical experiments on the intruder match rate are also conducted under various scenarios. In particular, we study (1) how best a template is divided into several small pieces called partial templates, each of which is to be stored in a separate silo; (2) how to do the matching by comparing partial templates in the locked-up silos, and letting only the results of these intra-silo comparisons be sent to the central tallying server for final scoring without requiring the whole templates in one location at any time.

키워드

Ising modelbiometric templatepartial templatedistributed biometricsIRISWATERMARKINGPROTECTIONPRIVACY
제목
Seeded Ising Model and Distributed Biometric Template Storage and Matching
저자
Choi, Hyeong InLee, SungjinMoon, Hwan PyoWee, Nam-SookKim, DaehoonKwon, Song-Hwa
DOI
10.3390/e23070849
발행일
2021-07
유형
Article
저널명
Entropy
23
7