Colour matching function learning

Luis Romero-Ortega*, Antonio Robles-Kelly

*Corresponding author for this work

    Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

    Abstract

    In this paper, we aim at learning the colour matching functions making use of hyperspectral and trichromatic imagery. The method presented here is quite general in nature, being data driven and devoid of constrained setups. Here, we adopt a probabilistic formulation so as to recover the colour matching functions directly from trichromatic and hyperspectral pixel pairs. To do this, we derive a log-likelihood function which is governed by both, the spectra-to-colour equivalence and a generative model for the colour matching functions. Cast into a probabilistic setting, we employ the EM algorithm for purposes of maximum a posteriori inference, where the M-step is effected making use of Levenberg-Marquardt optimisation. We present results on real-world data and provide a quantitative analysis based upon a colour calibration chart.

    Original languageEnglish
    Title of host publicationStructural, Syntactic, and Statistical Pattern Recognition - Joint IAPR International Workshop, SSPR and SPR 2012, Proceedings
    Pages318-326
    Number of pages9
    DOIs
    Publication statusPublished - 2012
    EventJoint IAPR International Workshops on Structural and Syntactic PatternRecognition, SSPR 2012 and Statistical Techniques in Pattern Recognition,SPR 2012 - Hiroshima, Japan
    Duration: 7 Nov 20129 Nov 2012

    Publication series

    NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
    Volume7626 LNCS
    ISSN (Print)0302-9743
    ISSN (Electronic)1611-3349

    Conference

    ConferenceJoint IAPR International Workshops on Structural and Syntactic PatternRecognition, SSPR 2012 and Statistical Techniques in Pattern Recognition,SPR 2012
    Country/TerritoryJapan
    CityHiroshima
    Period7/11/129/11/12

    Fingerprint

    Dive into the research topics of 'Colour matching function learning'. Together they form a unique fingerprint.

    Cite this