1. Overview of the Biometrics Lab Learning | Engagement | Discovery Biometric Standards, Performance, and Assurance Laboratory | Purdue University www.bspalabs.org www.twitter.com/bspalabs www.slideshare.net/bspalabs www.linkedin.com/companies/bspa-labs
2. Applied Biometrics The BSPA Laboratory was established in 2001 to meet the growing demand for applied research facilities in biometrics, primarily testing and evaluation Mission: To excel in the applied research of biometric technologies with a continued commitment to education and innovative research, as well as engaging academia and industry in all of our activities.
3. BSPA Lab Functional Areas Over 10 years of experience in education: undergraduate, graduate, distance, and industrial settings.
11. Learning Undergraduate and Graduate classes Biometric Technologies certificate CBP training Graduate certificate in Biometrics* Graduate courses Offered both online and on-campus
12. Undergraduate Course Automatic Identification and Data Capture Covers technology from bar codes, card technologies, and biometrics No prerequisites Offered online and offline
14. AIDC for the Enterprise Graduate level course in Automatic Identification Technologies Bar codes Biometrics RFID Card technologies
15. ICT Standards 1 credit course Repeatable 3 times Follows work in INCITS M1 and ISO/IEC JTC 1 SC37 Provides opportunities for students to provide technical comments Closely linked to biometric standards work
16. IT 55500 Biometric Technology Test Design, Performance, and Evaluation An introduction of methods of designing biometric testing, performance, and evaluation analyses. Methods of evaluating fingerprint, face, iris, and voice recognition data is explored using ROC curves, CMC, Rank statistics, and DET curves. Examines testing requirements from submission of IRB documents to the final analysis. Includes a component of comparative analysis within modalities.
17. IT 54000 Biometric Performance and Usability Analysis An introduction of test methodologies from disciplines outside of biometrics, which include: usability, ergonomics, human factors, human-computer interaction, Demonstrate how biometric data analysis can benefit from understanding how humans interact with biometric sensors during the testing and evaluation of biometric systems. Explores test methods, case studies, and prior biometric testing reports in order to develop a test methodology that includes information on how users interact with biometric systems.
18. IT 65700 Fingerprint Performance and Usability Covers topics of fingerprint capture, fingerprint feature extraction, fingerprint matching, and attacks on fingerprint systems. Requires analysis of real fingerprint data and the integration of fingerprint recognition in existing infrastructures. Development of a fingerprint recognition system is required.
19. IT 65800 Biometric Systems Interoperability: Applications and Challenges Provides a technology neutral approach to the discussion of biometric system interoperability. Examines the issues of biometric sub-systems of different biometric modalities and sub-systems of the general biometric model. Students will be able to critically evaluate the impact of interoperability of sub-systems on the performance of the entire system.
20. IT 54500 Biometric Technology and Applications Foundation course Six modules cover the IEEE CBP Body of Knowledge Biometric Fundamentals Biometric Modalities Biometric System Design and Evaluation Biometric Standards Social, Cultural and Legal Implications Biometric Applications
21. Access to courses Online – start at a time that is convenient to you Learn at your own pace For more information: http://www.bspalabs.org/about-us/contact-page/
26. Our Approach – Developing and Contribution to Standards Blind (2008)
27. Understanding Error The development of our approach consists of: Testing Research Standards Education
28. Research Agenda for Fall 2010 Biometric Operator Performance Face Recognition and the Indiana Department of Correction Standard Compliance of Legacy Biometric Data Modeling biometric modalities onto the HBSI (Human Biometric Sensor Interaction) method Understanding Biometric Error Habituation….
29. Biometric Operator Performance Analyzing the impact of different instructional methods of training on a biometric data collection agent and how would that affect the biometric transaction times during operational environment. Biometric modality: Mobile Iris Recognition Two methods of training: audio recording instruction soundless video instruction Two types of data collection agent from learning styles perspective: Verbal learner (prefers written and spoken explanations) Visual learner (prefers visual representations such as pictures, diagrams, flow charts)
30. Face Recognition and the Indiana Department of Correction Working with the Indiana Department of Correction to: Review and Analyze Current and Legacy Mug Shots Review current Mug Shot Capture Process Utilize Analysis of mug shots and review of capture process to: Propose an Optimized Capture Process Capture mug shots with more consistency that are standard compliant Implement Proposed Capture Process Analyze Mug Shots from Proposed Capture Process Determine if Proposed Process Successfully Optimized for Standard Compliance
31. Standard Compliance of Legacy Biometric Data Over time, government agencies collect a wide variety of face and fingerprint images from individuals. Historically, this data was collected manually, and stored in a filing cabinet, or scanned into a digital format. As agencies implement digital capture technologies, a question remains: what to do with the old data?
32. Standard Compliance of Legacy Biometric Data In this research project, we are analyzing face photographs that have been stored on paper, to examine whether these images are standard compliant. This three year project examined over 48,000 digital and paper-based photographs from the Indiana Department of Correction, with the intent to develop a list of recommendations on how to deal with legacy data. In this academic year, the results of the standard compliance analysis will be published, and available on the website.
33. Modeling biometric modalities onto the HBSI (Human Biometric Sensor Interaction) method The goal of this research is to provide the biometrics community with a comparative evaluation method for biometric devices that uses ergonomics, usability, biometric image quality, and traditional system performance criteria to evaluate the design and functionality of biometric devices and systems. This model was initially developed using fingerprint recognition as a base modality, but as the model matures, we have started to map the other modalities onto the model. This academic year will see hand geometry and iris recognition mapped for model validation.
34. Understanding Biometric Error Historically, biometric performance has relied on basic metrics such as FNMR and FMR, as well as Failure to Enroll, Failure to Acquire etc. As biometric deployments become widespread, and the number of people enrolled is in the millions, a 1% error rate is a significantly large number. A large part of our research portfolio is trying to understand this error, and providing new definitions and metrics. The goal of this research is to improve operational performance, design better systems (in line with the HBSI model), and to further the research communities understanding of biometric error.
35. Habituation…. Inside the biometric community, the definition of the word habituation varies from person to person. The general concept of the word implies that something happens as the user repeatedly uses a system. What this something is and the duration of repeatedly are the concern of this research.
36. Habituation… Some argue that the number of errors committed by the user will decrease as the user becomes more comfortable with the system, others argue that once comfortable with the system, the user’s interaction will become sloppy, therefore increasing the number of errors occurring. Others would like to quantify habituation as involving performance or image quality as opposed to error rates. No matter which side of the fence a researcher sits, there is a need to either accept an overarching definition of habituation or discard the word for a framework of variables
37. Habituation… Industrial engineering has long used the term habituation, but has an accepted definition of the word. This research will look at the definitions and changes in the use of the word habituation from both fields over time. Additionally, this study will try to create a framework in an attempt to create a cohesive model for the concept coined habituation. To do this, the study will utilize previous research and findings from multiple fields and sources.
39. Engagement | Outreach Mission Part of our mission is to engage and outreach to the biometrics community Successful deployment of biometrics is crucial to the growth of the biometrics industry Engage with companies to test, evaluate biometric devices
40. Engagement | Outreach Mission Engage with industry to solve complex problems Contribute to standards activities, both in the U.S. and internationally
42. Opportunities for students Of those that graduate from the lab 100% placement Work for government, contractors, private sector firms Many students have internships
44. How to contact us Knoy Hall of Technology, Purdue University 401 N. Grant Street West Lafayette, IN 47907-2021 www.bspalabs.org (765) 495-2311 contact@bspalabs.org