8 Fingerprint Individuality

  • 格式:pdf
  • 大小:801.03 KB
  • 文档页数:30

8 Fingerprint Individuality1

8.1 Introduction Expert testimony based on forensic evidence (such as handwriting, fingerprint, hair, bite marks, etc.) is routinely collected and presented in courtrooms. Among the various sources of evidence, fingerprints have been used in courts of law for almost 100 years and the testimony based on fingerprints carries substantial credibility and weight. The use of fingerprint evidence involves comparing salient features of a latent print lifted from a crime scene with those of rolled (full) impressions taken either from the known defendant or the entire criminal data-base. A reasonably high degree of match between the salient features leads the latent experts to testify irrefutably that the owner of the latent print and the defendant are one and the same person. For decades, the testimonies provided by latent print experts were almost never ex-cluded from these cases, and on cross-examination, the foundations and basis of such testimo-

fingerprint evidence is the assumption of discernible uniqueness; salient features of different individuals are observably different, and therefore, when two prints share many common fea-tures, the experts conclude that the owner of the two different prints is one and the same per-son. The assumption of discernible uniqueness (Saks and Koehler, 2005), although lacking sound theoretical and empirical foundations, allows forensic experts to offer unquestionable proof towards the defendant’s guilt, and to make matters worse, these experts are never ques-tioned on the uncertainty associated with their testimonials (that is, how frequently would an observable match between a pair of prints lead to errors in the identification of individuals; see

error rates which would be known from collecting population samples, analyzing the inherent feature variability, and reporting the corresponding probability of two different persons shar-ing a set of common fingerprint features.

1 Portions reprinted with permission from IEEE Transactions on Pattern Analysis and Machine Intelli-

gence, vol. 24, no. 8, pp. 1010−1025, 2002. © 2002 IEEE.

D. Maltoni et al., Handbook of Fingerprint Recognition, 341–370.

nies were rarely questioned (Cole (2001a, b)). Central to establishing an identity based on Haber and Haber (2004)). Thus, discernible uniqueness precludes the opportunity to establish

Springer-Verlag London Limited 2009 © 8 Fingerprint Individuality 342

A significant break from this trend occurred in the 1993 case of Daubert v. Merrell Dow Pharmaceuticals (113 S. Ct. 2786) where the U.S. Supreme Court ruled that in order for expert forensic testimony to be allowed in a court case, it had to be subject to three main criteria of scientific validation, that is, whether the particular tool or methodology in question (i) has been tested, (ii) has been subjected to peer-review, and (iii) possesses known error rates. Fol-lowing Daubert, fingerprint identification was first challenged in the 1999 case of U.S. v. Byron Mitchell (Criminal Action No. 96-407, US District Court for the Eastern District of Pennsylvania) under the premise that the uniqueness of fingerprints has not been objectively

of U.S. v. Byron Mitchell, fingerprint based identification has been challenged in more than

Cole (2006) has compiled a list of 22 known exposed cases of erroneous fingerprint identifica-tions made by forensic experts. In December 2005, the Massachusetts Supreme Judicial Court barred key fingerprint evi-

Court, case number K06-0545), judge Susan Souder ruled to exclude fingerprint evidence “be-cause the State did not prove in this case that opinion testimony by experts regarding the ACE-V (Analysis, Comparison, Evaluation, and Verification) method of latent print identifi-cation rests on a reliable factual foundation as required by MD Rule 5-702”. In making this ruling, the judge heavily relied on the case of an Oregon lawyer who was mistakenly linked through fingerprint analysis to the 2004 Madrid train bombings. These court rulings demon-strate both the awareness and the need to develop measures that reflect the confidence in a match when fingerprints are used as evidence in the courts of law. Fingerprint individuality deals with the problem of quantifying the extent of uniqueness of a fingerprint. How similar should two fingerprints be before we can conclude with high confidence that they are from the same finger? What are the measures of fingerprint individuality that reflect the extent of un-certainty in the observed match? The main challenge in studying fingerprint individuality is to develop statistical models that adequately describe the variability of fingerprint features in a target population. These models can, in turn, be used to derive the probability of a random match between two different fingerprints picked arbitrarily from the target population. Eliciting candidate models for repre-senting the variability of fingerprint features is not an easy task due to the complex nature of this variability. Candidate models should satisfy two important requirements, namely, (i) flexi-bility, that is, the models can represent a wide range of distributional characteristics of finger-print features in the population, and (ii) associated confidence measures can be easily obtained from these models. The fingerprint individuality problem can be formulated in many different ways, depend-ing on which one of the following aspects of the problem is under examination: (i) the indi-viduality problem may be cast as determining the probability that any two or more individuals