Individual Evaluation: The Biometric Menagerie

When one considers how widely the appearance of a biometric can vary, along with all the idiosyncratic behaviors that govern a user’s interaction with a system, the hypothesis that all members of a population perform equally seems optimistic. However, this assumption is implicit in much of the analysis of the previous chapter, which focused on a high-level, aggregated statistics. There is no doubt that these performance measures are necessary, but it is also important to recognize that they do not tell the whole story. A system is comprised of data from a variety of sources, so collective statistics like ROC and CMC graphs necessarily preclude some information. Within any given biometric system, it is almost always the case that some users perform better than others, and it is these individual differences that are the subject of this chapter.

Section 7.1.2 introduced the concept of match score distributions for genuine and impostor matches. These distribution functions give the probabilities of genuine and impostor matches over the range of possible scores. It is a central theme of this book that these distributions are the basis for all performance analysis, and many, if not all, aspects of biometrics system have a distribution-based interpretation. For example, in Sect. it was demonstrated that false match and false non-match rates are related to the degree of overlap between the genuine and impostor distributions. Similarly, in Sect. the impact of database size on identification performance was illustrated through shifting rank 1 impostor score distributions. Once again, the key concept of this chapter can be posed of in terms of match score distributions.


Biometric System Impostor Performance Capture Zone Iris Recognition False Match 


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. 1.
    Adler, A., Youmaran, R., Loyka, S.: Towards a measure of biometric feature information.In Press: Pattern Analysis and Applications (2008)Google Scholar
  2. 2.
    Authenti-Corp: IRIS06 draft final report.
  3. 3.
    Daniel, W.: Applied Nonparametric Statistics. Wadsworth Publishing Company (1989)Google Scholar
  4. 4.
    Doddington, G., Liggett, W., Martin, A., Przybocki, M., Reynolds, D.: Sheep, goats, lambs and wolves a statistical analysis of speaker performance in the NIST 1998 speaker recognition evaluation. In: Proceedings of ICSLP-98 (1998)Google Scholar
  5. 5.
    Hicklin, A., Watson, C., Ulery, B.: The myth of goats: How many people have fingerprints that are hard to match? Tech. Rep. NIST IR 7271, National Institute of Standards and Technology (2005)Google Scholar
  6. 6.
    Maio, D., Maltoni, D., Cappelli, R., Wayman, J.L., Jain, A.K.: FVC2002: Second fingerprint verification competition.In: Proceedings of ICPR, vol. 3, pp. 811–814 (2002)Google Scholar
  7. 7.
    Wayman, J.L.: Multi-finger penetration rate and ROCvariability for automatic fingerprint identification systems.Tech. rep., National Biometric Test Center (1999)Google Scholar
  8. 8.
    Wittman, M., Davis, P., Flynn, P.: Empirical studies of the existence of the biometric menagerie in the FRGC 2.0 Color Image Corpus. In: Proceedings of CVPRW (2006)Google Scholar
  9. 9.
    Yager, N., Amin, A.: Fingerprint verification using two stage optimization. Pattern Recognition Letters 27, 317–324 (2006)CrossRefGoogle Scholar
  10. 10.
    Yager, N., Dunstone, T.: The biometric menagerie. Submittedto IEEE PAMI (2007)Google Scholar
  11. 11.
    Yager, N., Dunstone, T.: Worms, chameleons, phantoms and doves: New additions to the biometric menagerie. In: Proceedings of AutoID (2007)Google Scholar

Copyright information

© Springer-Verlag US 2009

Personalised recommendations