Illumination discrimination for chromatically biased illuminations:

implications for colour constancy

Supplemental Material

Stacey Aston, Ana Radonjić, David H. Brainard and Anya C. Hurlbert

Instructions pdf

Comparison illumination lookup tables
Tables specifying the difference between the target and comparison illuminations in ΔE L*u*v* units are provided as tab delimited text. We used three different lookup tables (as explained in the paper). For each table, each row corresponds to one nominal illumination step (shown in column 1). The reference illumination condition is indicated by the header label in the first row; illumination direction within a condition is indicated by the header label in the second row.

Fixed white-point lookup tables (txt) were constructed from the illumination spectra XYZ tristimulus values; XYZ tristimulus coordinates of the neutral reference illumination [in xyY notation: 0.31, 0.33, 50] are used as the white point (for illumination spectra to XYZ to L*u*v* conversion).

Image mean lookup tables (txt) were constructed using the same white point as the fixed XYZ lookup tables but using mean image chromaticity under each illumination (and not the illumination spectra) for conversion to CIELUV coordinates. 

Variable white point lookup tables (txt) were constructed from the illumination spectra XYZ tristimulus values. For each reference illumination condition, it uses that reference illumination's XYZ values as the white point for conversion of that condition’s test illuminations to CIELUV coordinates (in xyY notation these were: Neutral reference illumination: [0.31, 0.33, 50], Blue reference illumination: [0.26, 0.26, 50], Green reference illumination: [0.30,0.38, 50], Red reference illumination: [0.33,0.26, 50], Yellow reference illumination: [0.40, 0.39, 50]).

Illuminations. For each reference illumination condition we provide spectral power distribution for the target and comparison illuminations as tab delimited text. Spectra are provided in W/m²/sr. Files are grouped by illumination. 1st column shows wavelength in nm (380 - 780 nm, 1 nm steps). Target illumination is shown in the 2nd column in each file. For the remaining columns, header labels correspond to comparison illumination spectra with numbers denoting the nominal comparison illumination step (difference relative to the target). 

Neutral reference condition: Blue Yellow Green Red

Blue reference condition: Blue Yellow Green Red

Green reference condition: Blue Yellow Green Red

Red reference condition: Blue Yellow Green Red

Yellow reference condition: Blue Yellow Green Red

Thresholds. The table provides individual participants’ thresholds for each reference illumination condition as tab delimited text. (txt)
Each row corresponds to one subject (subject id numbers are assigned arbitrarily) and shows measured thresholds for the four illumination directions in each of the five reference illumination conditions (20 threshold per participant). The reference illumination condition is indicated by the header label in the first row; illumination direction within a condition is indicated by the header label in the second row. Thresholds are computed from raw data following the methods described in the Data Analysis section of the paper.

Additional statistical analyses

Post-hoc tests. For completeness, we also provide the results of post-hoc tests for 1) pairwise comparison of different illumination-change directions for each reference illumination condition and 2) pairwise comparison of different reference illuminations for each illumination-change direction condition. The results are provided for analyses based on each of the three sets of lookup tables (pdf).