Skip to main content

Robust Multi-Biometrics Identification Using Soft-Biometric Intelligence

Researcher: Marios Savvides

Research Area: Security of Cyber-Physical Systems

Abstract

Scope: Sponsored by the DoD Biometrics Task Force, the goal of this project is to develop a robust and intelligent fusion method for multi-biometric identification. Current systems treat face recognition and black boxes. We aim to add intelligence to the system matching by building robust soft-biometric classifiers that can determine soft-biometric traits such as gender, ethnicity, presence of moustache/beard, glasses, tattoos or other unique features or occlusions, which can lead to poor score matching. For example, two facial images may provide a low score but near the threshold of a successful match however, one image the person has a beard and the other does not. Soft-Biometrics can fuse the intelligence of knowing that one face has a beard and the other does not, so that even if the score is border-line, this suggests that it could be a successful match and the lower score is due to image variability (e.g., the person shaved his beard). Currently, human analysts perform this kind of intelligence, and thus we aim to automatic this process for more intelligent matching and fusion process.

Outcomes: A complete system that will be able to analyze a facial image and provide a wide series of soft-biometric feature analysis (e.g. gender, detection of facial hair and location, presence of glasses, etc) and a fusion scheme to be integrated to current face matching technology.