Project Summary 

• $800,000 in Federally-funded research into the application of multimodal fusion systems and high-resolution fingerprint technologies in justice applications 

IBG Responsibilities  

• Design test methodology for biometric data collection and processing
• Implement six commercial fingerprint, face, and iris recognition matchers to process archived biometric images
• Collect and process high-resolution fingerprints through customized fingerprint capture system
• Place results and findings in context of justice applications such as criminal identification and mobile biometric data collection

In 2005, IBG was awarded two NIJ grants totaling $900,000 to study cutting-edge biometric technologies for criminal justice applications.

Cutting-Edge Fingerprint Algorithm Research

IBG conducted an 11-month evaluation examining the utility of Level III fingerprint features found in high-resolution fingerprint images - ranging from 2000dpi to 3791dpi - for automated identification. IBG studied whether scanning fingerprints at high resolutions generates friction ridge detail and discriminating information sufficient to support utilization of new automated matching tools.

IBG collected and evaluated over 25,000 high-resolution fingerprint images to study the  permanence and distinctiveness of Level III features such as pores and ridge morphology.

Multimodal Fusion Research

IBG conducted an 11-month research project, Efficient Field-Optimized Multimodal Biometric System (EFOMBS), that evaluated the accuracy of multimodal fusion and normalization techniques used in biometric systems.

Results provided guidance to law enforcement professionals determining how to most effectively leverage biometric data, as well as implement biometric matchers, in present-day and future biometric systems.

IBG processed its internal database of over 100,00 fingerprint, face, and iris samples through commercial matching engines to generate comparison scores. For each inter-modal matcher combination, comparison scores were normalized (placed on a common scale) and fused (combined though some statistical method).

Results indicate that, at most operational thresholds, certain multimodal techniques regularly generate stronger match scores than other techniques, regardless of the matcher combination. In most but not all cases, the best fusion technique was better than the strongest single matcher.