http://www.cs.sfu.ca/~colour/data/index.htmlHowever, the associated meta data is likely to be corrected/updated more frequently in this version of the interface.
Questions, comments, and problems with this data should be directed to
This directory contains some the data presented in:
Kobus Barnard, Lindsay Martin, Brian Funt, and Adam Coath, " Data for Colour Research," Color Research and Application, Volume 27, Issue 3, pp. 148-152, 2002.
(The appropriate archival reference for this data).
This data in this directory is used in the following publications:
Kobus Barnard and Brian Funt, " Camera characterization for color vision research," Color Research and Applications, Vol. 27., No. 3, pp. 153-164, 2002.
Chapter 3 of Kobus Barnard, "Practical Colour Constancy," Phd thesis, Simon Fraser University, School of Computing (1999)
Kobus Barnard and Brian Funt, " Camera calibration for color vision research," SPIE Conference on Electronic Imaging, Human Vision and Electronic Imaging IV, SPIE Vol. 3644, 1999, pages 576-585.
The data was largely collected by Lindsay Martin under the guidance of Kobus Barnard in Brian Funt's Computational Colour Vision Laboratory.
The main data set is file of 598 spectra and the corresponding RGB's for a Sony DXC-930 3 chip CCD video camera. This camera was used to collect much of the data available from this web site, and thus further exploration of its calibration may be of interest to some.
The measurements consist of 23 patches (all but black) of the Macbeth color chart under 26 different illuminants. The spectra consist of 101 measurements, from 380 to 780 nm, in 4nm increments. This is the output provided by a PhotoResearch 650 spectrometer. The camera is a sony DXC-930 3 chip CCD camera used with gamma off, and a color temparature of 3200K (indoor illumination). The camera has a significant output with the lens cap on. This output can be found in the camera black file below. All RGB measurements are the spatial average of a 50 by 50 area in the center of the frame. Flare was reduced as best as possible. Further details of the experimental setup are provided in the paper.
IMPORTANT: Some of the RGB's have the value 255.0. These data values are likely clipped and should not be used. We include all data, since most (or all) RGB have at least one non clipped value. In other words, even if the red channel is clipped, the green or blue channel may contain useful information.
598 RGB (598 lines of 3 numbers)
Camera black RGB (1 line of 3 numbers)
598 Spectra (598 spectra, each consisting of 101 lines, 1 number per line, gzipped).