we look at the results of voting and we look at the
IAI update by Joe Polski.
Fast Capture of Rolled-Equivalent Fingerprints and Palm Prints 4 new awards at NIJ
• Requirements represent a major step forward in finger and palm print capture technology: • Capture of 10 rolled-equivalent fingerprints in 15 seconds or less; • Capture of both palms in1 minute and less; • Meets or exceeds FBI & NIST specifications; • Reduces failure-to-enroll rates; • Produces a working device suitable for independent test in 18 months; and is • Affordable, rugged, portable, relatively unobtrusive in size, and deployable
in the near future. Awards were made to: Cross Match (Formerly
Smith’s Detection Inc.)
Flexible polymer plastic 2-dimensional sensor
array on a polymer foil base substrate. The Polymer foil is thin and elastic
and can be applied to any 1- dimensional curved surface designed for optimum
capture geometry of a non-rolled finger or palm.
TBS North America Inc.
Patent pending ‘Segment ImagingTM’ system is
based on touchless, optoelectronics technology using customized hardware and
software components. Software system includes algorithms for hardware
calibration and for real-time image quality control.
Carnegie
Mellon University
Proposed system constructs a visual
3-dimensional model of both hands, including palms, fingerprints,
fingertips, and sides of the fingers. • Overlay images from multiple cameras on a complete 3-D model of both hands using active models to normalize for pose; • Extract ridge detail based on contrast assessment under varying illumination effects; and,
• Translate 3-D images to standardized
formats.
Based on DARPA autonomous computing modules.
Segmentation algorithm based on Face Recognition PIE datasets.
University
of Kentucky (Funded Directly by DHS)
A calibration pattern will be positioned
inside the box, such that the cameras can automatically calibrate themselves
prior to each scan eliminating the need for operator assistance. A
structured light illumination system will acquire a 3-D surface scan of a
hand. Once the hand’s pose is determined, a flash-style projection source
will project a single structured-light pattern onto the hand with the
component cameras each acquiring a high-resolution, digital photograph.
Forensic Science Communications
NIJ’s FY 05 Friction Ridge Research Development Projects
1) Title: Quantitative Assessment of the Individuality of Friction Ridge Patterns Grantee: Research Foundation of SUNY Amount: $596,478.00 Abstract: The research proposes to increase understanding of the discriminative power of friction ridge patterns. Two tasks will be addressed: (i) assess existing statistical models of friction ridge individuality and propose new models for error rates, probability of match /exclusion and strength of evidence, e.g., the strength of a match can be expressed using newer statistical techniques developed in the allied forensic disciplines of DNA matching and speaker verification, and (ii) study the issue of quantity and quality of friction ridge data that need to be present for matching, e.g., how does the number and combinations of minutiae present affect individualization? In the first task different models of individuality and their assumptions will be compared, e.g., traditional models which lead to an assertion that the probability of duplication is a small x% and new probabilistic models based on distributions of similarity values conditioned on belonging to the same or different individual. The quantity-quality study will parameterize error rates on the basis of minutiae available and their combinations. Methods to be employed will be software-based and will include both existing algorithms for minutiae extraction/matching as well as newer algorithms for extracting fingerprint characteristics. Examples of the latter are: compound minutiae consisting of minutiae combinations, counting the numbers and combinations of minutiae that occur, extracting features not automatically extracted at present, etc. Resources to conduct empirical studies of friction ridge prints will include a recently constructed research database of friction ridge patterns collected from a population of twins– this database, prepared by latent print examiners contains ten prints, palm-prints, and latent prints, will also help determine relative similarity between identical twins and the general population. Deliverables will include software to quantify the strength of evidence and software for extracting and counting friction ridge characteristics. The work will be conducted with the guidance of latent print examiners at federal, state and local levels.
2) Title: Analysis of Level III Characteristics at High Resolutions Grantee: International Biometric Group, LLC Amount: $461,495.00 (Phase 1) Abstract: International Biometric Group (IBG), Aprilis Inc., and the Crime Scene Services Section of the Massachusetts State Police (MSP-CSSS) propose a research project that evaluates (1) the frequency and permanence of Level III characteristics and (2) tools that enable capture, processing, and statistical evaluation of these characteristics' quality and strength. The Project evaluates card-based and live-scanned fingerprint data at resolutions from 500dpi to 4000dpi, assessing the degree to which increased resolution enhances the utility of Level III characteristics. Through comparison of Level III characteristics derived from genuine and impostor populations, the Project generates data that describes the discriminating power of Level III characteristics. The Project further compares Level III matching results with Level II results to assess multimodal correlation and relative matching power. Expanding on existing research techniques – and utilizing Aprilis capture hardware and matching software – evaluation design encompasses the following:
1. Digitization of friction ridge data from several thousand tenprint cards, selected as paired records with temporal variation. 2. Processing and comparison of card-based fingerprints for location and segmentation of Level III characteristics, generating data on friction ridge permanence. 3. Collection, conversions, and digitization of multiple fingerprint impressions at 500dpi, 1000dpi, 2000dpi, and 4000dpi from a population of approximately 1000 test subjects (comprising approximately 150,000 samples). 4. Classification and regional sampling of friction ridge data to establish robustness within minimal ridge areas. 5. Cross-comparative processing of live-scan samples, generating data on ridge distinctiveness. 6. Analysis of Level III characteristic performance in conjunction with and relative to Level II characteristics. 7. Optimization of Aprilis software tools.
Study of Level III characteristics is timely as performance requirements for civil biometric systems (e.g. national ID, border management) are approaching the limits of Level II algorithms. The justice community has a vested interest in the continued collection of fingerprints for Civil Identification applications; as such databases have utility in law enforcement scenarios. This Project will help to quantify the discriminating power of Level III friction ridge data, support technology implementation decisions, and improve tools available for use in friction ridge applications.
3) Title: Adding Human Expertise to the Quantitative Analysis of Fingerprints Grantee: Indiana University Amount: $431,255.00 Abstract: Quantitative approaches to fingerprint identification rely on different approaches derived from minutiae detection, orientation computations and other sources of information. These approaches extract out relevant features that can be matched across prints. We propose toimprove upon these approaches by incorporating data derived from expert latent print examiners. We have developed a software tool that provides a very rich description of the expert's matching process, and we extract from this data elements of expertise that can be incorporated into quantitative analyses of fingerprints. We demonstrate how this data can be used to identify regions of interest using variations on support vector machines, how we can use the temporal information to identify the nature of the information used by experts, and how data reduction techniques reveal the fundamental features used by experts, which may be more than just minutiae. Finally, we integrate across different levels of a print to use the 'gist' or category of a print to improve the feature extraction of parts of the print. Our particular quantitative methods take advantage of our expert data, but the data derived from experts could be used to improve many different quantitative approaches, and we will make our data available to other researchers for use in their statistical models.
4) Title: Latent-Print Detection by Macro-Raman Imaging Grantee: Oak Ridge National Laboratory (Through and Interagency Agreement with DOE) Amount: $299,000.00 (Phase 1) Abstract: Fingerprints deposited on many surfaces often go undetected once latent prints age over a few hours, especially when exposed to UV radiation. The ability to develop latent fingerprints is often influenced by many factors including print-type (clean/eccrine through oily/sebaceous), humidity, light, surface matrix, etc. Recent findings on the fundamental chemistry of superglue fuming, a prominent method for developing prints on non-porous surfaces, revealed methods capable of enhancing the ability of develop latent fingerprints that would otherwise go undetected. In many cases, treatment of the print with vapor from 75% acetic acid dramatically enhanced development by superglue fuming. However, this enhancement was not effective on fingerprints exposed to UV radiation from sun or fluorescent lighting, especially on surfaces containing iron (III). Such surfaces would include firearms, knives, ammunition, automobiles, etc. In addition, the enhancement method is complex and not easily amenable to field applications. Thus, a real need exists to efficiently and effectively detect latent fingerprints on all surfaces regardless of the print type or environmental exposure factors. To accomplish this goal, further study is needed to better characterize constituents and associated degradation products originating from fingerprint secretions deposited on a range of matrices. Through an understanding of time-related changes in fingerprint components, discrimination between fingerprint constituents and the deposition surface is expected to facilitate the development of enhanced friction ridge visualization methods. Simplistic methods that increase the detection sensitivity for macro-Raman Imaging will be targeted. With this type of discriminatory power, an increase in the average print area and quality, as well as in the differentiation between fresh and aged prints, are expected.
In order to exploit the anticipated findings, researchers from the Oak Ridge National Laboratory (ORNL) are teaming with experts in chemical imaging – ChemImage, Corp. Currently, ChemImage is developing a Raman-based imaging system under a TSWG initiative. The goal this proposal is to gain a better understanding of fingerprint and fingerprint degradation chemistry, employ methods to enhance Raman-based latent-print visualization, and utilize the enhancement methods to modify the ChemImage system for field applications. Such a system would enhance the efficiency and quality of latent detection in cases involving WMD events (non-contact print detection), assault, murder, etc.
The tasks necessary to achieve the stated goal are complex and require a unique combination of specialists who understand fingerprint decomposition, chemical enhancement, chemical imaging, and imaging enhancement. ORNL proposes to team with ChemImage to develop a macro-Raman chemical imaging methodology to accomplish the stated objective.
5) Title: Quantifying the Dermatoglyphic Growth Patterns in Children through Adolescence Grantee: Ultra Scan Corporation Amount: $126,601.00 Abstarct: Ultra-Scan Corporation, a pioneer in the use of ultrasound for livescan fingerprint imaging with over 15 years experience in researching, commercializing, integrating, and deploying fingerprint identification systems, proposes to contribute to the scientific body of friction ridge structure knowledge. To date, this body of knowledge does not provide scientific understanding regarding the growth pattern of fingerprints to enable the positive identification of children over a period of several years.
Ultra-Scan will develop a predictive model that is a 2-way mapping (from younger to older, and older to younger) of the minutiae location, both spatial and orientation, of two sets of fingerprints captured at different points in time. The predictive model will be based on a control group of several hundred children ranging from age 2 to 18, repeatedly imaging their fingerprint ridge structure over the course of five years.
The establishment of a 2-way fingerprint predictive growth model will assist latent examiners in the identification of young children, improve the accuracy of automated latent search engines, and expand the scientific body of knowledge regarding fingerprint patterns. Ultra-Scan will disseminate the resulting data analysis to the broader forensic science community in addition to the NIJ, through peer review publications and technical presentations. Ultra-Scan will leverage expertise and existing equipment to deliver statistically significant findings at minimal cost.
6) Title: Improving Methods for Fingerprint Development on Hand-Guns Grantee: Israeli National Police (Through an Interagency Agreement with TSWG) Amount: $75,000.00 Abstract: This project will involve two stages of research. In the first stage, the factors that affect the life, durability, and recovery of fingerprints on hand-guns will be studied. After gaining a better understanding of what happens to the fingerprints, the second stage will attempt to develop more successful methods for fingerprint development on hand-guns.
The major portion of this second stage will involve research into optimization of the cyanoacrylate method. Today, this is the method of choice for processing hand-guns and throughout the years, this method has been optimized in regards to development conditions and methods. However, the cyanoacrylate used was always either ethyl or methyl. Today, there are seven or eight types of cyanoacrylates available (butyl, octyl, etc.). Little research has been done into choosing and using the best cyanoacrylate type for the type of surface being processed. Just as different types of cyanoacrylate have to be adapted to the material being glued, the same may hold true for fingerprint processing.
|
||||||||||||||||||||||||||||||||||||||||||||||||||||
