Biometric system based on kinect skeletal, facial and vocal features

Yaron Lavi, Dror Birnbaum, Or Shabaty, Gaddi Blumrosen

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

4 Scopus citations

Abstract

Identification of human subject in different environments plays a significant role in many fields like security and health care. The identification can be performed by using different sensory metrics, often named “biometric”. Traditional biometric technologies are based mainly on fingerprint, retina, voice, and face. In this study, the spontaneous use of skeleton, facial, and vocal metrics is being investigated. For this, a Microsoft Kinect (“Kinect”) system, which was mainly built to estimate human subject kinematic features are deployed. Kinect is affordable, non-wearable, and has the potential to assess joints location, voice, and facial properties simultaneously. A set of skeletal, facial, and vocal features is extracted, and create a “Kinect Signature” that is used to identify different subjects in the scene. The methods were verified by a set of four experiments simulating common realistic scenarios. The experiments indicate that the skeleton, facial, and vocal metrics derived from the Kinect can differentiate between different subjects. The results of this work indicate that while skeletal based metrics are usually more accessible compared to facial and vocal metrics, facial and vocal metrics are more accurate. Aggregation of all data streams improves biometric system performance and their continuity in different environments and times. Such systems can be a base for an affordable, accurate real-time biometric system, that can be deployed at home, and public facilities like hospitals.

Original languageEnglish
Title of host publicationProceedings of the Future Technologies Conference (FTC) 2018 - Volume 1
EditorsRahul Bhatia, Kohei Arai, Supriya Kapoor
PublisherSpringer Verlag
Pages884-903
Number of pages20
ISBN (Print)9783030026851
DOIs
StatePublished - 2019
Externally publishedYes
EventFuture Technologies Conference, FTC 2018 - Vancouver, BC, Canada
Duration: 15 Nov 201816 Nov 2018

Publication series

NameAdvances in Intelligent Systems and Computing
Volume880
ISSN (Print)2194-5357

Conference

ConferenceFuture Technologies Conference, FTC 2018
Country/TerritoryCanada
CityVancouver, BC
Period15/11/1816/11/18

Bibliographical note

Publisher Copyright:
© Springer Nature Switzerland AG 2019.

Keywords

  • Biometric
  • Face recognition
  • Kinect
  • Posture
  • Voice recognition

Fingerprint

Dive into the research topics of 'Biometric system based on kinect skeletal, facial and vocal features'. Together they form a unique fingerprint.

Cite this