Type of Publication: Article in Collected Edition

HotFoot: Foot-Based User Identification Using Thermal Imaging

Author(s):

Saad, Alia; Izadi, Kian; Ahmad Khan, Anam; Knierim, Pascal; Schneegass, Stefan; Alt, Florian; Abdelrahman, Yomna

Title of Anthology:

Proceedings of the 2023 CHI Conference on Human Factors in Computing Systems

Publisher:

Association for Computing Machinery

Location(s):

New York, NY, USA

Publication Date:

2023

ISBN:

9781450394215

Keywords:

User identification, Thermal Imaging, Footprint

Digital Object Identifier (DOI):

doi:10.1145/3544548.3580924

Citation:

Download BibTeX

Abstract

We propose a novel method for seamlessly identifying users by combining thermal and visible feet features. While it is known that users’ feet have unique characteristics, these have so far been underutilized for biometric identification, as observing those features often requires the removal of shoes and socks. As thermal cameras are becoming ubiquitous, we foresee a new form of identification, using feet features and heat traces to reconstruct the footprint even while wearing shoes or socks. We collected a dataset of users’ feet (N = 21), wearing three types of footwear (personal shoes, standard shoes, and socks) on three floor types (carpet, laminate, and linoleum). By combining visual and thermal features, an AUC between 91.1% and 98.9%, depending on floor type and shoe type can be achieved, with personal shoes on linoleum floor performing best. Our findings demonstrate the potential of thermal imaging for continuous and unobtrusive user identification.