Face liveliness detection based on texture and color features

Li Song, Hongbin Ma*

*Corresponding author for this work

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

8 Citations (Scopus)

Abstract

Nowadays, face recognition has been used in many security occasions, but few of them have ability to distinguish real and fake faces. Besides, many researches on face liveliness detection mainly focused on intrusive methods, which are not user-friendly in practice. This paper proposes a novel non-intrusive face liveliness detection method based on the analysis of texture and color features. More specifically, this method adopts an improved local ternary pattern (LTP) to classify the nearby pixels. Based on the face pixel analysis, the infinity norm of pixel matrices is added as new features. The effectiveness of feature selection has been validated by different kinds of experiments on three challenging face anti-spoofing databases (NUAA, CASIA FASD and Replay-attack). This method reaches a compromise between number of features and accuracy, which means it also works on embedded systems.

Original languageEnglish
Title of host publication2019 IEEE 4th International Conference on Cloud Computing and Big Data Analytics, ICCCBDA 2019
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages418-422
Number of pages5
ISBN (Electronic)9781728114095
DOIs
Publication statusPublished - Apr 2019
Event4th IEEE International Conference on Cloud Computing and Big Data Analytics, ICCCBDA 2019 - Chengdu, China
Duration: 12 Apr 201915 Apr 2019

Publication series

Name2019 IEEE 4th International Conference on Cloud Computing and Big Data Analytics, ICCCBDA 2019

Conference

Conference4th IEEE International Conference on Cloud Computing and Big Data Analytics, ICCCBDA 2019
Country/TerritoryChina
CityChengdu
Period12/04/1915/04/19

Keywords

  • LBP
  • LTP
  • color moment
  • infinity norm
  • liveliness detection

Fingerprint

Dive into the research topics of 'Face liveliness detection based on texture and color features'. Together they form a unique fingerprint.

Cite this