A General Weighted Multi-scale Method for Improving LBP for Face Recognition

X Wei, H. Wang, Gongde Guo, Huan Wan

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

7 Citations (Scopus)

Abstract

LBP (Local Binary Pattern) is a popular image descriptor (feature) that has been widely used in face recognition. LBP has some parameters, and different parameter values leads to different LBP feature vectors. In practice usually only one feature vector is used for one image, thus information about image content is not utilised fully by LBP. In this paper a novel way of utilising LBP features more fully is presented. Different LBP feature vectors are extracted for one image, corresponding to different combinations of LBP parameter values. These vectors are weighted and used in a distance function. Then the k-nearest neighbour classifier is used. Experiments have been conducted on the AR database. Results show this method does indeed produce better classification performance, suggesting that more information considered this way can have values.
Original languageEnglish
Title of host publicationUbiquitous Computing and Ambient Intelligence. Personalisation and User Adapted Services
Subtitle of host publicationUCAmI 2014
Place of PublicationSwitzerland
PublisherSpringer International Publishing
Pages532-539
Number of pages8
Volume8867
ISBN (Electronic)978-3-319-13102-3
ISBN (Print)978-3-319-13101-6
DOIs
Publication statusPublished (in print/issue) - 2014
EventUbiquitous Computing and Ambient Intelligence - Belfast, United Kingdom
Duration: 2 Dec 20145 Dec 2014

Publication series

NameLecture Notes in Computer Science
PublisherSpringer Nature
Volume8867

Conference

ConferenceUbiquitous Computing and Ambient Intelligence
Abbreviated titleUCamI 2014
Country/TerritoryUnited Kingdom
CityBelfast
Period2/12/145/12/14

Keywords

  • Multi-scale LBP
  • Face Recognition

Fingerprint

Dive into the research topics of 'A General Weighted Multi-scale Method for Improving LBP for Face Recognition'. Together they form a unique fingerprint.

Cite this