2018 | The Year in Review – MULTISPECTRAL BIOMETRICS – NOVEL BIOMETRICS – VIDEO ANALYTICS – BEHAVIORAL BIOMETRICS – CREDIBILITY ASSESSMENT – TRUST & PRIVACY – SCIENCE OF BIOMETRICS – CYBERSECURITY – MOBILE & COMPUTING – FUSION & PERFORMANCE
A NATIONAL SCIENCE FOUNDATION INDUSTRY/UNIVERSITY COOPERATIVE RESEARCH CENTER
C I Te R |
OUTREACH | WORKSHOPS CHALLENGE PROBLEM
A YEAR IN REVIEW
With our interdisciplinary group of faculty, researchers and students, CITeR is the only National Science Foundation (NSF) Industry/University Cooperative Research Center (I/UCRC) focusing on serving its affiliates in the rapidly growing areas of identity and biometrics.
CITeR Affiliates
Outreach/Workshops
Aetna Health Inc.
CITeR Student Workshop — Forty attendees participated in a student workshop held in conjunction with the Fall 2018 Program Review in Niagara Falls, New York. The student workshop included learning sessions in DNA, deep learning and privacy and was followed by a student research poster session.
Army Futures Command — Combat Capabilities Development Command — Armaments Center (CCDC — Armaments) AWARE Inc.
40
Defence Research and Development Canada (DRDC)
master’s and PhD students currently engaged in CITeR research
Defense Forensic Science Center (DFSC) Department of Defense — Defense Forensics and Biometrics Agency (DFBA) Department of Homeland Security — Office of Biometric Identity
25
Students work in breakout sessions on deep learning techniques.
Outreach — In the spring of 2018, CITeR hosted 50 students from the Atlanta Public Schools for an interactive affiliate panel review session followed by a session with hands-on stations where the students talked with CITeR graduate students about biometrics research.
Management (OBIM)
undergraduates currently engaged in CITeR research
Department of Homeland Security — Science and Technology Directorate (S&T)
Challenge Problem/Affiliate Focus
Federal Bureau of Investigation (FBI) InCadence Strategic Solutions
15
faculty engaged in research and teaching biometrics
Infineon Technologies AG Integrated Biometrics Precise Biometrics Northrop Grumman Corporation Qualcomm Incorporated
15
webinars featuring results from recently completed CITeR projects
SRC Inc.
CITeR hosts yearly Challenge Problem workshops, which typically highlight an affiliate or related organization and allow for problem/ solution discussion between CITeR researchers and affiliate organizations. The Spring 2018 Challenge Problem Workshop was held at the Defense Forensic Science Center, Forest Park, Georgia. Previous workshops have been held at the NYPD Lower Manhattan Security Coordination Center, New York City; Peace Bridge, Buffalo, New York; and FBI Biometric Center of Excellence, Clarksburg, West Virginia.
Students talk with CITeR graduate students at biometric learning stations in Atlanta, Georgia.
StereoVision Imaging Inc. Veridium ZOLOZ
3
new affiliates added in 2018
c i t e r. c l a r k s o n . e d u
Attendees of the Spring 2018 Challenge Problem Workshop.
C I Te R | 1
RESEARCH HIGHLIGHTS
RESEARCH HIGHLIGHTS
TECHNOLOGY TRANSFER
- WEST VIRGINIA UNIVERSITY
- CLARKSON UNIVERSITY
- WEST VIRGINIA UNIVERSITY
- UNIVERSITY AT BUFFALO
Latent Fingerprint Distortion Correction and Reconstruction CITeR graduate student, Ali Dabouei, and research advisors, Dr. Nasser Nasrabadi and Dr. Jeremy Dawson, at West Virginia University focused on latent fingerprint distortion and reconstruction software to improve latent fingerprint automated recognition. Their research received two awards: the Best Student Paper Award at Biometrics: Theory, Applications and Systems and the International Association for Pattern Recognition Best Biometrics Student Paper Award at the International Conference on Biometrics, both in 2018.
Examples of the fingerprint reconstructions on real latent fingerprints generated by CITeR researchers at West Virginia University. For each fingerprint image, the corresponding latent sample and the reconstructed sample are demonstrated.
Elastic distortion of fingerprints negatively affects the performance of fingerprint recognition systems. Also, in a negative recognition scenario in which users may intentionally distort their fingerprints, this distortion will prevent the recognition system from identifying malicious users. Current methods aimed at addressing this problem have limitations. They are often inaccurate because they estimate distortion parameters based on the ridge frequency map and orientation map of input samples. A rectification model was developed based on a Deep Convolutional Neural Network (DCNN) to accurately estimate distortion parameters from the input image. Using a comprehensive database of synthetic distorted samples, the DCNN learns to accurately estimate distortion bases 10 times faster than the dictionary search methods used in previous approaches. Evaluating the proposed method on public databases of distorted samples shows that it can significantly improve the matching performance of distorted samples.
Physiologically Accurate Fingerprint Phantom
Prototype of a state-of-the-art test phantom finger developed by CITeR researchers at the University of Buffalo.
c i t e r. c l a r k s o n . e d u
CITeR graduate student, Phil Schneider, and research advisor, Dr. Kwang Oh, at University at Buffalo, The State University of New York, focused on creating a physiologically accurate phantom finger for the testing, development and characterization of an ultrasonic fingerprint system. As smartphone vendors continue to develop mobile biometric technologies, they drive the key performance indicators associated with system performance, such as acquisition time, FAR/FRR and liveness detection. The created state-of-the-art test phantom finger provides a realistic, stable, repeatable representation of biological/physiological features such as fingerprints, vascular patterns, blood flow, dermatographic features, tissue and the static and dynamic touch forces of fingers. This piece of test metrology allows for the quantification and analysis of biometric system performance, both at a superficial-imaging and subdermal level.
GPU-Accelerated Facial Video Analysis CITeR graduate student, Xuan Qi, and research advisors, Dr. Chen Liu and Dr. Stephanie Schuckers, at Clarkson University focus on the need for rapid facial recognition in video for national and industrial security. Facial recognition in video requires processing huge amount of video streams using video analytics. Performing real-time face tracking on live streaming video or on a large video repository already poses a considerable computational challenge, considering that a regular video stream from surveillance and security cameras has 30 frames per second (FPS) or more. Adding facial recognition would definitely complicate the situation and require tremendous computing power locally or a cloud computing platform, for example. To address these challenges, CITeR researchers developed an innovative key frame analysis algorithm, which tracks face(s) across frames and automatically selects the frame with the best face(s) quality based on carefully selected metrics. In addition, the researchers also optimized the algorithm, using a state-of-theart graphic processing unit (GPU) as the hardware computing engine. A GPU is traditionally used for computer video display, but it performs parallel computation better than traditional CPU computation. This quality-based video facial biometric software has been installed at the FBI, a CITeR affiliate, and an evaluation by a second affiliate is in progress.
Technology Transfer — Transition to DHS OBIM Software from CITeR researcher Guodong Guo was transitioned to DHS OBIM for evaluation. The software, which was developed as part of the CITeR project “Understandable Face Image Quality Assessment,” automatically assesses face image quality, even in challenging real-world photos. Face image quality assessment is important because the quality can vary significantly due to different imaging sensors, compression techniques, video frames and/or image acquisition conditions. Assessing face image quality automatically, quickly and precisely in real-world images is very challenging, and the software addresses these challenges.
C I Te R | 3
FINANCIALS | OUTCOMES
GRADUATE HIGHLIGHTS | AWARDED RESEARCH PROJECTS
CITeR Income Summary 2014-18 $1,200,000
West Virginia University Michael Martin, BS in Computer Engineering, BS in Biometric Systems, 2012; PhD in Computer Engineering, expected 2020. Hired by CITeR affiliate Northrop Grumman as a systems architect and biometric subject-matter expert.
Michael Martin
“I have worked on several CITeR projects related to facial recognition. The most recent involves using enhancements on passport images to improve facial recognition capabilities. Previous projects involved using relatively low-cost thermal cameras to determine the temperature in regions of the face. ... CITeR allowed me to network with affiliates and make connections with industry professionals that led to great opportunities with my current employer. My CITeR funded projects taught me invaluable skills and provided knowledge in many areas important to industry and government agencies.”
University at Buffalo
Phil Schneider, PhD in Electrical Engineering, 2018. Hired by ACV Auctions in the R&D division to head up technology innovations for vehicle identification and conditioning.
Phil Schneider
“We have coined the phrase ‘vehicle fingerprint’ as we are constantly looking through massive amounts of data to identify a vehicle (i.e., make, model, VIN). I implement some of the same tools and practices seen throughout CITeR, such as ML, computer vision, DLN, phantom models, etc. ... CITeR provided me with opportunities to learn, network, fund my research and ultimately grow into a scientist, landing me in my current role.”
Clarkson University Nick Short, BS in Electrical Engineering, 2016. Hired by CITeR affiliate SRC Inc. as a systems engineer supporting electronic warfare projects. “During my time with CITeR, I had the unique opportunity as an underclassman to develop and assume full responsibility for a bio-radar research project. Through this research project, I gained valuable technical knowledge, such as radar theory and MATLAB coding skills, that I would not have otherwise developed through my normal slate of coursework. Additionally, I was responsible for planning the necessary steps to get from the project’s inception to its ultimate completion.” Nick Short
c i t e r. c l a r k s o n . e d u
CITeR Fall 2017 Projects – Cross-Device Forensic Speaker Verification Using Coupled Deep Neural Networks, $50K. – Deep Fingerprint Matching From Non-Contact to 2-D Legacy Rolled Fingerprints, $50K. – Deep Hashing for Secure Multimodal Biometrics, $50K. – Developing an Automated Method to Remove Labeling Noise in Very Large Scale Dataset, $50K. – Incorporating Biological Models for Iris Authentication in Mobile Environments (Phase II), $50K. – Learning High Entropy Robust Features for Privacy Preserving Facial Templates, $50K. – Light-Weight Machine Learning for Biometric Tasks on IoT Devices, $50K. – “Liveness Detection”: Photoacoustic Imaging of Mechanically Accurate Test Phantom Finger, $50K. – A Practical Evaluation of Free-Text Keystroke Dynamics, $42K.
CITeR Spring 2018 Projects – – – – – –
Multistage Fusion of Biometric Matchers, $40K. Face Anti-Spoofing: A Comprehensive Evaluation, $40K. From DNA to Face: Deducing Facial Morphology From Human Genomic Data, $35K. Evaluation of Speaker Recognition Solutions to Guide Prototype Development, $50K. Enabling Secure and Privacy Preserving Authentication via Blockchain/Smart Contract and Biometrics, $75K. Development and Validation of Radar-Based Biometric Recognition (Gateway Project), $40K. – Biometric Aging in Children — Phase II, $60K.
$40K
affiliate membership gains access to the community and a portfolio of in projects
$675K
NSF I/UCRC
$1,000,000
Affiliate $
$800,000 $600,000 $400,000 $200,000 $0
2014
2015
2016
2017
2018
CITeR Publications
Year
total # of publications (all researchers)
2016
50
2017
54
2018
57
CITeR Citations
Year
total # of citations (all researchers)
2016
3362
2017
3242
2018
3245
Database Requests Year
# fulfilled requests (all universities)
2016
75
2017
55
2018
57
BOOKS
Surveillance in Action, P. Karampelas, and T. Bourlai (Eds.), Springer, 2018.
C I Te R | 5
citer.clarkson.edu Contacts
Affiliate Advisory Board Chair
Clarkson University Dr. Stephanie Schuckers Director 315-268-6536 sschucke@clarkson.edu
Laura Holsopple Managing Director 315-268-2134 lholsopp@clarkson.edu
West Virginia University Dr. Matthew Valenti Site Director 304-293-9139 matthew.valenti@mail.wvu.edu
Dr. Nasser Nasrabadi Site Co-Director 304-293-4815 nasser.nasrabadi@mail.wvu.edu
University at Buffalo Dr. Venu Govindaraju Site Director 716-645-3321 venu@cubs.buffalo.edu
Srirangaraj Setlur Site Co-Director 716-645-1568 setlur@buffalo.edu
Michigan State University (planned) Dr. Arun Ross 517-353-9731 rossarun@cse.msu.edu
Kody West Technical Fellow Northrop Grumman Corporation Robert J. DelZoppo (Outgoing Affiliate Advisory Board Chair) Assistant Vice President Strategic Technology Programs SRC, Inc. Affiliate Executive Committee Terry Riopka Director of Research AWARE Inc. William Zimmerman Chief Engineer Defense Forensics and Biometrics Agency Chris Chamberlin Project Manager Office of Biometric Identity Management) Department of Homeland Security Reza Derakhshani Chief Scientist ZOLOZ
c i t e r. c l a r k s o n . e d u