An improved method for Daugman's iris localization algorithm

Comput Biol Med. 2008 Jan;38(1):111-5. doi: 10.1016/j.compbiomed.2007.07.007. Epub 2007 Sep 27.

Abstract

Computer-based automatic recognition of persons for security reasons is highly desirable. Iris patterns provide an opportunity for separation of individuals to an extent that would avoid false positives and negatives. The current standard for this science is Daugman's iris localization algorithm. Part of the time required for analysis and comparison with other images relates to eyelid and eyelash positioning and length. We sought to remove the upper and lower eyelids and eyelashes to determine if separation of individuals could still be attained. Our experiments suggest separation can be achieved as effectively and more quickly by removing distracting and variable features while retaining enough stable factors in the iris to enable accurate identification.

Publication types

  • Research Support, Non-U.S. Gov't

MeSH terms

  • Algorithms*
  • Eyelashes / anatomy & histology
  • Eyelids / anatomy & histology
  • Humans
  • Image Enhancement / methods*
  • Iris / anatomy & histology*
  • Pattern Recognition, Automated / methods*
  • Reproducibility of Results