Presentation 2005/3/21
Color constancy as Bayesian inference on scene statistics
Toshihiro TOYOTA, Hidekyuki HONJYO, Shigeki NAKAUCHI,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) The problem of colour constancy is formulated as Bayesian inference on scene statistics to recover the colour of an incident illumination. We begin with the data rendering equation describing the relation between the illuminant and statistics of surface colours, and statistics of the observed scene. We here focus on the first and second-order statistics (mean and luminance-colour correlation) of the scene which are likely to be cues for the illuminant colour as suggested by recent psychophysical observations. Then, we construct prior distributions for statistics of the illuminants and a set of surface colours as the probability density describing particular illuminants and surfaces existing in the world. Simulation results show that Bayesian estimates for the illuminant colour which maximize a posteriori probability computed with a given scene are robust across hue distribution changes of surface colours.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) color constancy / scene statistics / Bayesian inference
Paper # NC2004-158
Date of Issue

Conference Information
Committee NC
Conference Date 2005/3/21(1days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair
Vice Chair
Secretary
Assistant

Paper Information
Registration To Neurocomputing (NC)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Color constancy as Bayesian inference on scene statistics
Sub Title (in English)
Keyword(1) color constancy
Keyword(2) scene statistics
Keyword(3) Bayesian inference
1st Author's Name Toshihiro TOYOTA
1st Author's Affiliation Depertment of Computer Sciences, Toyohashi University of Technology()
2nd Author's Name Hidekyuki HONJYO
2nd Author's Affiliation Depertment of Computer Sciences, Toyohashi University of Technology
3rd Author's Name Shigeki NAKAUCHI
3rd Author's Affiliation Depertment of Computer Sciences, Toyohashi University of Technology
Date 2005/3/21
Paper # NC2004-158
Volume (vol) vol.104
Number (no) 758
Page pp.pp.-
#Pages 6
Date of Issue