Upper‐Bounding the Incidence Rate of Associations Between Camouflage Uniforms and Surveillance Images*

Abstract: Camouflage garments can be associated with surveillance images of a crime scene even in the absence of unique wear marks or very high‐quality images. However, the probability of an accidental association, or incidence rate, is significant. The present work describes and validates a method for estimating the incidence rate based on a statistical model of the garment manufacturing process. The model was developed primarily for use with the current U.S. Army Combat Uniform (ACU), but can be applied to any camouflage garment. Eight garment manufacturers were studied, and all sources of variation in the manufacturing process were characterized. The marking and spreading procedures were found to be dominant and consistent sources of variation. However, some sources of variation, in particular those because of human operators, were not consistent enough to accurately characterize. Sources of variation that could not be well‐characterized were ignored in the statistical model, yielding a worst‐case estimate that is an upper‐bound to the true incidence rate. The model was evaluated for a variety of cases. Depending on the quality of the surveillance image, the manufacturing parameters, and the local population, incidence rates range from about 3% to negligibly small. The model was validated by returning to one manufacturer, and sampling a large number of completed garments and estimating empirical match probabilities. The empirical probabilities validated the estimates of the worst‐case incidence rate and also demonstrated that typical incidence rates are significantly lower.     

Computation of likelihood ratios in fingerprint identification for configurations of any number of minutiae

Recent court challenges have highlighted the need for statistical research on fingerprint identification. This paper proposes a model for computing likelihood ratios (LRs) to assess the evidential value of comparisons with any number of minutiae. The model considers minutiae type, direction and relative spatial relationships. It expands on previous work on three minutiae by adopting a spatial modeling using radial triangulation and a probabilistic distortion model for assessing the numerator of the LR. The model has been tested on a sample of 686 ulnar loops and 204 arches. Features vectors used for statistical analysis have been obtained following a preprocessing step based on Gabor filtering and image processing to extract minutiae data. The metric used to assess similarity between two feature vectors is based on an Euclidean distance measure. Tippett plots and rates of misleading evidence have been used as performance indicators of the model. The model has shown encouraging behavior with low rates of misleading evidence and a LR power of the model increasing significantly with the number of minutiae. The LRs that it provides are highly indicative of identity of source on a significant proportion of cases, even when considering configurations with few minutiae. In contrast with previous research, the model, in addition to minutia type and direction, incorporates spatial relationships of minutiae without introducing probabilistic independence assumptions. The model also accounts for finger distortion.     

Computation of likelihood ratios in fingerprint identification for configurations of three minutiae

Recent challenges to fingerprint evidence have brought forward the need for peer-reviewed scientific publications to support the evidential value assessment of fingerprint. This paper proposes some research directions to gather statistical knowledge of the within-source and between-sources variability of configurations of three minutiae on fingermarks and fingerprints. This paper proposes the use of the likelihood ratio (LR) approach to assess the value of fingerprint evidence. The model explores the statistical contribution of configurations of three minutiae using Tippett plots and related measures to assess the quality of the system. Features vectors used for statistical analysis have been obtained following a preprocessing step based on Gabor filtering and image processing to extract minutia position, type, and direction. Spatial relationships have been coded using Delaunay triangulation. The metric, used to assess similarity between two feature vectors is based on an Euclidean distance measure. The within-source variability has been estimated using a sample of 216 fingerprints from four fingers (two donors). Between-sources variability takes advantage of a database of 818 ulnar loops from randomly selected males. The results show that the data-driven approach adopted here is robust. The magnitude of LRs obtained under the prosecution and defense propositions stresses upon the major evidential contribution that small portions of fingermark, containing three minutiae, can provide regardless of its position on the general pattern.     

Introducing a semi-automatic method to simulate large numbers of forensic fingermarks for research on fingerprint identification

Statistical research on fingerprint identification and the testing of automated fingerprint identification system (AFIS) performances require large numbers of forensic fingermarks. These fingermarks are rarely available. This study presents a semi-automatic method to create simulated fingermarks in large quantities that model minutiae features or images of forensic fingermarks. This method takes into account several aspects contributing to the variability of forensic fingermarks such as the number of minutiae, the finger region, and the elastic deformation of the skin. To investigate the applicability of the simulated fingermarks, fingermarks have been simulated with 5-12 minutiae originating from different finger regions for six fingers. An AFIS matching algorithm was used to obtain similarity scores for comparisons between the minutiae configurations of fingerprints and the minutiae configurations of simulated and forensic fingermarks. The results showed similar scores for both types of fingermarks suggesting that the simulated fingermarks are good substitutes for forensic fingermarks.     

Estimation and evidence in forensic anthropology: age-at-death

A great deal has previously been written about the use of skeletal morphological changes in estimating ages-at-death. This article looks in particular at the pubic symphysis, as it was historically one of the first regions to be described in the literature on age estimation. Despite the lengthy history, the value of the pubic symphysis in estimating ages and in providing evidence for putative identifications remains unclear. This lack of clarity primarily stems from the fact that rather ad hoc statistical methods have been applied in previous studies. This article presents a statistical analysis of a large data set (n = 1766) of pubic symphyseal scores from multiple contexts, including anatomical collections, war dead, and victims of genocide. The emphasis is in finding statistical methods that will have the correct "coverage."Coverage" means that if a method has a stated coverage of 50%, then approximately 50% of the individuals in a particular pubic symphyseal stage should have ages that are between the stated age limits, and that approximately 25% should be below the bottom age limit and 25% above the top age limit. In a number of applications it is shown that if an appropriate prior age-at-death distribution is used, then "transition analysis" will provide accurate "coverages," while percentile methods, range methods, and means (+/-standard deviations) will not. Even in cases where there are significant differences in the mean ages-to-transition between populations, the effects on the stated age limits for particular "coverages" are minimal. As a consequence, more emphasis needs to be placed on collecting data on age changes in large samples, rather than focusing on the possibility of inter-population variation in rates of aging.     

Resolving paternity relationships using X-chromosome STRs and Bayesian networks

X-chromosomal short tandem repeats (X-STRs) are very useful in complex paternity cases because they are inherited by male and female offspring in different ways. They complement autosomal STRs (as-STRs) allowing higher paternity probabilities to be attained. These probabilities are expressed in a likelihood ratio (LR). The formulae needed to calculate LR depend on the genotype combinations of suspected pedigrees. LR can also be obtained by the use of Bayesian networks (BNs). These are graphical representations of real situations that can be used to easily calculate complex probabilities. In the present work, two BNs are presented, which are designed to derive LRs for half-sisters/half-sisters and mother/daughter/paternal grandmother relationships. These networks were validated against known formulae and show themselves to be useful in other suspect pedigree situations than those for which they were developed. The BNs were applied in two paternity cases. The application of the mother/daughter/paternal grandmother BN highlighted the complementary value of X-STRs to as-STRs. The same case evaluated without the mother underlined that missing information tends to be conservative if the alleged father is the biological father and otherwise nonconservative. The half-sisters case shows a limitation of statistical interpretations in regard to high allelic frequencies.     

A test for carrion fly full siblings: a tool for detecting postmortem relocation of a corpse

We propose a genetic test for full sibship for a pair of carrion flies that could reveal the postmortem relocation of a corpse. A carrion fly larva is sometimes left behind when a corpse is moved. The discovery of full sibling larvae of approximately the same developmental stage at two locations would strongly suggest that a corpse was moved between those two sites. Distributions of pairwise comparisons of relatedness (R) coefficients were generated using amplified fragment length polymorphism profiles for nine samples of laboratory-generated full siblings as well as for a reference sample of nonfull sibling Phormia regina (Diptera: Calliphoridae). The mean relative R coefficient, a pairwise measure of the proportion of shared alleles, was 0.479 (±0.289 SD) for full siblings, close to the theoretical expectation of 0.5. A likelihood ratio (LR) test was based on observed distributions of R. R >0.55 corresponded to an LR >1000 favoring full sibship for that pair of individuals.     

On using gait in forensic biometrics

Given the continuing advances in gait biometrics, it appears prudent to investigate the translation of these techniques for forensic use. We address the question as to the confidence that might be given between any two such measurements. We use the locations of ankle, knee, and hip to derive a measure of the match between walking subjects in image sequences. The Instantaneous Posture Match algorithm, using Harr templates, kinematics, and anthropomorphic knowledge is used to determine their location. This is demonstrated using real CCTV recorded at Gatwick International Airport, laboratory images from the multiview CASIA-B data set, and an example of real scene of crime video. To access the measurement confidence, we study the mean intra- and inter-match scores as a function of database size. These measures converge to constant and separate values, indicating that the match measure derived from individual comparisons is considerably smaller than the average match measure from a population.     

Prediction of eye color from genetic data using Bayesian approach

Prediction of visible traits from genetic data in certain forensic cases may provide important information that can speed up the process of investigation. Research that has been conducted on the genetics of pigmentation has revealed polymorphisms that explain a significant proportion of the variation observed in human iris color. Here, on the basis of genetic data for the six most relevant eye color predictors, two alternative Bayesian network model variants were developed and evaluated for their accuracy in prediction of eye color. The first model assumed eye color to be categorized into blue, brown, green, and hazel, while the second variant assumed a simplified classification with two states: light and dark. It was found that particularly high accuracy was obtained for the second model, and this proved that reliable differentiation between light and dark irises is possible based on analysis of six single nucleotide polymorphisms and a Bayesian procedure of evidence interpretation.     

Assessment of approximate likelihood ratios from continuous distributions: a case study of digital camera identification

A reported likelihood ratio for the value of evidence is very often a point estimate based on various types of reference data. When presented in court, such frequentist likelihood ratio gets a higher scientific value if it is accompanied by an error bound. This becomes particularly important when the magnitude of the likelihood ratio is modest and thus is giving less support for the forwarded proposition. Here, we investigate methods for error bound estimation for the specific case of digital camera identification. The underlying probability distributions are continuous and previously proposed models for those are used, but the derived methodology is otherwise general. Both asymptotic and resampling distributions are applied in combination with different types of point estimators. The results show that resampling is preferable for assessment based on asymptotic distributions. Further, assessment of parametric estimators is superior to evaluation of kernel estimators when background data are limited.     

Handwriting evidence evaluation based on the shape of characters: application of multivariate likelihood ratios

A novel Bayesian methodology has been developed to quantitatively assess handwriting evidence by means of a likelihood ratio (LR) designed for multivariate data. This methodology is presented and its applicability is shown through a simulated case of a threatening anonymous text where a suspect is apprehended. The shape of handwritten characters a, d, o, and q of the threatening text was compared with characters of the true writer, and then with two other writers, one with similar and one with dissimilar characters shape compared to the true writer. In each of these three situations, 100 draws of characters were made and the resulting distributions of LR were established to consider the natural handwriting variation. LR values supported the correct hypothesis in every case. This original Bayesian methodology provides a coherent and rigorous tool for the assessment of handwriting evidence, contributing undoubtedly to integrate the field of handwriting examination into science.