Proceedings of the IEEK Conference (대한전자공학회:학술대회논문집)

The Institute of Electronics and Information Engineers (대한전자공학회)
권호 표지

A Hybrid Nonsmooth Nonnegative Matrix Factorization for face representation

다양한 얼굴 표현을 위한 하이브리드 nsNMF 방법

  • Lee, Sung-Joo (School of Electrical and Electronic Engineering, Yonsei University) ;
  • Park, Kang-Ryoung (Department of Electronics Engineering, Dongguk University) ;
  • Kim, Jai-Hie (School of Electrical and Electronic Engineering, Yonsei University)
  • 이성주 (연세대학교 전기전자 공학과) ;
  • 박강령 (동국대학교 전자공학과) ;
  • 김재희 (연세대학교 전기전자 공학과)
  • Published : 2008.06.18
Download PDF
⟨ Previous Next ⟩

Abstract

The human facial appearances vary globally and locally according to identity, pose, illumination, and expression variations. In this paper, we propose a hybrid-nonsmooth nonnegative matrix factorization (hybrid-nsNMF) based appearance model to represent various facial appearances which vary globally and locally. Instead of using single smooth matrix in nsNMF, we used two different smooth matrixes and combine them to extract global and local basis at the same time.

Keywords