A Study of Digital Camera Colorimetric Characterization Based on Polynomial Modeling

GW Hong,MR Luo,PA Rhodes
DOI: https://doi.org/10.1002/1520-6378(200102)26:1<76::aid-col8>3.0.co;2-3
2000-01-01
Abstract:The digital camera is a powerful tool to capture images for use in imageprocessing and colour communication. However, the RGB signals generated by adigital camera are device-dependent, i.e. different digital cameras produce differentRGB responses for the same scene. Furthermore, they are not colorimetric, i.e. theoutput RGB signals do not directly correspond to the device-independent tristimulusvalues based on the CIE standard colorimetric observer. One approach for deriving acolorimetric mapping between camera RGB signals and CIE tristimulus values usespolynomial modelling and is described here. The least-squares fitting technique wasused to derive the coefficients of 3× n polynomial transfer matrices yielding a modellingaccuracy typically averaging 1 Δ E units in CMC(1:1) when a 3× 11 matrix is used.Experiments were carried out to investigate the repeatability of the digitising system,characterisation performance when different polynomials were used, modellingaccuracy when 8-bit and 12-bit RGB data were used for characterisation and the numberof reference samples needed to achieve a reasonable degree of modelling accuracy.Choice of characterisation target and media and their effect on metamerism have beenexamined. It is demonstrated that a model is dependent upon both media and colorantand applying a model to other media/colorants can lead to serious eye-camerametamerism problems.
What problem does this paper attempt to address?