Cristin-resultat-ID: 789076
Sist endret: 18. oktober 2016, 10:20
NVI-rapporteringsår: 2006
Resultat
Vitenskapelig artikkel
2006

Investigating von Kries-like adaptation using local linear models

Bidragsytere:
  • Graham Finlayson
  • Steve Hordley og
  • Ali Alsam

Tidsskrift

Color Research and Application
ISSN 0361-2317
e-ISSN 1520-6378
NVI-nivå 1

Om resultatet

Vitenskapelig artikkel
Publiseringsår: 2006
Volum: 31

Importkilder

ForskDok-ID: r07003633

Beskrivelse Beskrivelse

Tittel

Investigating von Kries-like adaptation using local linear models

Sammendrag

In this article we explore the conditions under which a von Kries-like model of color constancy is exact. This investigation is motivated by the fact that in practice the model has been shown to work-well for a wide range of imaging devices despite the fact that previous analysis of the necessary and sufficient conditions for the model to work predicts that it should perform poorly for real lights and surfaces. We present a modified theory, that reconciles this apparent contradiction and that is based on the observation that von Kries-like adaptation treats sensor responses independently of one another. Starting front this point we show how to recover,for a single sensor, set of surfaces, and reference illuminant, the set of von Kries illuminants: all lights for which von Kries-like adaptation is a perfect model of illumination change. To help us in this task we use a linear model of surface reflectance, but importantly, we use a local model, that is, a model derived by looking at reflectance only in the region to which the sensor of interest is sensitive. Adopting such a model and treating sensors independently of one another, we show that our new theory accurately predicts the good practical performance of a von Kries-like model. (c) 2006 Wiley Periodicals, Inc.

Bidragsytere

Graham Finlayson

  • Tilknyttet:
    Forfatter
    ved Norges teknisk-naturvitenskapelige universitet

Steve Hordley

  • Tilknyttet:
    Forfatter
  • Tilknyttet:
    Forfatter
    ved University of East Anglia

Ali Alsam

  • Tilknyttet:
    Forfatter
    ved Institutt for datateknologi og informatikk ved Norges teknisk-naturvitenskapelige universitet
1 - 3 av 3