USDOE Human Subjects Research Database, fiscal year 1995

Sandia National Laboratories

Project Identifier: SNL-95-05

Project Title:

Evaluation of the Performance of the Facial Image Biometric Identifiers

Principle Investigator: Mr. Frank Bouchier

Project started in: 1995

Fiscal Year 1995 Funding for Research on Human Subjects:

Project Funding Information:
Project received funding in Fiscal Year 1995.
Project used human subjects in Fiscal Year 1995.

Funding Sources:

DOE: Office of Safeguards and Security

Amount: $19,000

Information on Use of Human Subjects:

Project does not involve use of multiple protocols/subprojects.

IRB Review:
Type of Review: Expedited
Most Recent Approval: April 25, 1995

Number of Human Subjects in the Last Reporting Period for this Project: 52
(Reporting periods vary.)

Type of Human Subjects Involvement:

Instrument/Device/Product Testing or Man-Machine Studies:

Use of human subjects to develop/test instruments, materials, devices, or objects.

(a. Objectives, b. Methodology, c. Ionizing Radiation, Radioactive Substances, or Chemical Substances to which human subjects are exposed, d. Involvement of Human Subjects [d.1. procedures used, d.2. risks if any])

Prototype devices to be evaluated use the unique characteristic of individual facial images for biometric identification. Both these systems acquire digital images of people's faces, and verification of identity is accomplished by matching the image with a reference image obtained from a previous enrollment. Characterizing the performance of these devices in terms of error rates, length of time required for verification, user acceptability, vulnerability to attack, and identification of categories of people who might not be able to use these devices is the purpose of this test.

Volunteers will be trained in the use of the devices and will use them as often as possible until a suitable database of transactions is obtained. The number of false rejects will be compared to the number of transactions to establish the false reject rate. Then volunteers' templates will be deleted and each volunteer will try the devices several times. Error rates (for example, false matches) will be calculated, and observations during the test and user surveys following the test will address the other performance considerations.

None of the published results will contain individual performance or identification of any kind. There are no known risks associated with the use of the biometric identification devices to be tested, and there are no known risks to any of the study procedures.