Illumination Modeling and Normalization for Face Recognition

  • Authors:
  • Haitao Wang;Stan Z. Li;Yangsheng Wang;Weiwei Zhang

  • Affiliations:
  • -;-;-;-

  • Venue:
  • AMFG '03 Proceedings of the IEEE International Workshop on Analysis and Modeling of Faces and Gestures
  • Year:
  • 2003

Quantified Score

Hi-index 0.00

Visualization

Abstract

In this paper, we present a general framework for face modelingunder varying lighting conditions. First, we show that a facelighting subspace can be constructed based on three or moretraining face images illuminated by non-coplanar lights. Thelighting of any face image can be represented as a point in thissubspace. Second, weshow that the extreme rays, i.e. the boundaryof an illumination cone, cover the entire light sphere. Therefore,a relatively sparsely sampled face images can be used to build aface model instead of calculating each extremely illuminated faceimage. Third, we present a face normalization algorithm,illumination alignment, i.e. changing the lighting of one faceimage to that of another face image. Experiments are presented.