CITeR’s Technology Readiness Assessment Center (TRAC)

CITeR is a National Science Foundation (NSF) Industry/University Cooperative Research Center (I/UCRC) focused on the advancement of automated human biometrics and credibility assessment security vulnerabilities and countermeasures, multimodal systems, biometric image coding and quality, performance and modeling frameworks, and socio-legal studies. The Center functions as a discovery and innovation cooperative between academia, industry, and government. CITeR has executed over 80 proof-of-concept projects tightly coupled to industry and government needs.

While CITeR’s role is invaluable, TRAC puts in place a structure to more effectively bridge the gap between promising project outcomes achieved through small-scale IUCRC proof-of concept funding and the more extensive assessment of those outcomes for their viability for transition to private and government sector systems. CITeR management solicited and received input from affiliates regarding the past projects perceived to have the potential for technology transition. Selected projects fell into three categories: Biometric Surveillance, Forensic Biometrics, and Fusion and System Assessment. Each project has been assigned its readiness assessment champions, i.e., the experts from industry and government organizations. These individuals will monitor the progress on each project and offer these studies directions on how to best assess their maturity and application / transition potential. Selected projects are discussed briefly below.

Unconstrained Face Recognition Under Non-Ideal Conditions – this project seeks to improve the ability to restore low quality facial images for comparison against higher quality images by using algorithms designed to improve images. The improvement uses distinctive facial marks to identify particular individuals. The researchers plan to use significantly degraded passport photographs that have been faxed, compressed, downscaled, and printed and then use the designed algorithm to improve the photograph to a useful level of quality.

Macro Features In Iris and Matching – this project seeks to improve the ability to manually mark macro features of iris images, as well as allowing for those the user to then either add to or delete those changes. The project will attempt to improve iris matching capabilities based upon those same macro features. After accomplishing both of those outcomes, it will then seek to create the capability to automatically mark macro images with manual editing to occur afterward. Finally, it will allow for the automatic retrieval of images from databases based upon specific macro features.

Retrieving Face Images Based On Facial Marks – this project seeks to further improve recognition accuracy by augmenting COTSrecognition system with face mark information. To test its progress, the project will conduct a large scale data experiment conducive of real operational gallery databases to better understand the usefulness of face marks in recognition. This project will also use high resolution face images to perform texture analysis at the micro-dermal level as a form of mark based recognition. Finally, it will perform a cost-benefit analysis of the developed techniques against currently used methodologies.

Establishing The Chain Of Evidence for Digital Biometric Files – the project seeks to establish forensic accountability to biometric information collections by concentrating on fingerprinting of fingerprint sensors, soft biometric authentication, and biometric watermarking techniques for change tracking. This project will advance current techniques by improving upon device fingerprinting algorithms and will analyze how keystroke dynamics can enhance and monitor user authentication. This project will feature the use of an open source watermarking technique for system demonstration.

Joint Modeling of Biometric System Performance, Risk, and Fusion Source – this project will develop analytical performance models of biometric security applications, explore the specific security risks associated with biometric misclassifications, and analyze fusion algorithms to both minimize risks and improve performance.