Illumination Normalization for Robust Face Recognition Against Varying Lighting Conditions

  • Authors:
  • Shiguang Shan;Wen Gao;Bo Cao;Debin Zhao

  • 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

Evaluations of the state-of-the-art of both academic facerecognition algorithms and commercial systems have shown thatrecognition performance of most current technologies degrades dueto the variations of illumination. This paper investigates severalillumination normalization methods and proposes some novelsolutions. The main contribution of this paper includes: (1)A GammaIntensity Correction (GIC) method is proposed to normalize theoverall image intensity at the given illumination level; (2) ARegion-based strategy combining GIC and the Histogram Equalization(HE) is proposed to further eliminate the side-lighting effect; (3)A Quotient Illumination Relighting (QIR) method is presentedtosynthesize images under a pre-defined normal lighting conditionfrom the provided face images captured undernon-normal lightingcondition. These methods are evaluated and compared on the Yaleillumination face database B and Harvard illumination facedatabase. Considerable improvements are observed. Some conclusionsare given at last.