An anatomy of IrisCode for precise phase representation

Adams Kong, Dapeng Zhang, Mohamed Kamel

Research output: Journal article publicationConference articleAcademic researchpeer-review

3 Citations (Scopus)


IrisCode, a widely deployed iris recognition algorithm, developed in 1993 and continuously modified by Daugman has attracted considerable attentions. IrisCode using a coarse phase representation has number of properties such as rapid matching, binomial imposter distribution and predictable false acceptance rate. Although many similar coding methods have been developed for irises and palmprints based on IrisCode, a theoretical analysis of IrisCode has not been provided. In this paper, we aim at studying (1) the nature of IrisCode, (2) the property of the phase of Gabor function, (3) the extension of bitwise hamming distance and (4) the theoretical foundation of the binomial imposter distribution and extending the coarse phase representation to a precise phase representation. Precisely, we demonstrate that IrisCode is a clustering algorithm with four prototypes; the locus of a Gabor function is a two-dimensional ellipse with respect to the phase parameter and bitwise hamming can be regarded as angular distance. Using these properties, we provide a precise phase representation for IrisCode with an effective implementation for filtering and matching. Practically, the imposter distribution of IrisCode follows binomial distribution. However, the theoretical evidence is incomplete according to our analysis.
Original languageEnglish
Article number1699870
Pages (from-to)429-432
Number of pages4
JournalProceedings - International Conference on Pattern Recognition
Publication statusPublished - 1 Dec 2006
Event18th International Conference on Pattern Recognition, ICPR 2006 - Hong Kong, Hong Kong
Duration: 20 Aug 200624 Aug 2006

ASJC Scopus subject areas

  • Computer Vision and Pattern Recognition


Dive into the research topics of 'An anatomy of IrisCode for precise phase representation'. Together they form a unique fingerprint.

Cite this