Defining color calibration information

For this example, we assumed that your display is pretty standard (and like some of ours). These assumptions are not going to be precisely correct for your CRTs; they are not even close for printers or LCD displays.

A broad description of the issues concerning color calibration may be found in the following references:

• D. H. Brainard -- Color Research and Application
• B. A. Wandell -- Foundations of Vision

---

1. Effects of the display primaries on the human cones

   To compute the effect of the display primaries on the cones, we need to know the spectral power distribution (SPD) of the display, and the relative absorptions of the human cones.

   We include a file containing the display SPDs of one of our monitors in the file displaySPD.mat. The file contains the variables:

   • wavelength: a vector describing the wavelengths of the cone sensitivities (370:73))
   • displaySPD: a 361x3 matrix of the spectral power distributions of the red, green, and blue primaries

   We usually compute with respect to the human cone estimates from Smith and Pokorny. These are represented in the file named SmithPokornyCones.mat. The file contains the following variables:

   • wavelength: a vector describing the wavelengths of the cone sensitivities (370:73))
   • cones: a 361 x 3 matrix matrix describing the cone sensitivities at the corresponding wavelengths

   The display SPD and cone sensitivities are loaded in as:

   load displaySPD
   load SmithPokornyCones
   

   The cone sensitivities and the display spectral power distributions we have assumed are plotted below:

   plot(wavelength,cones);
   plot(wavelength, displaySPD);
   

   Sample monitor SPD plot	Cone sensitivity plot

   Now, we can compute the 3 x 3 transformation that maps the linear intensity of the display r,g,b signals into the cone absorptions (l,m,s).

   rgb2lms = cones'* displaySPD;

---

2. Gamma correction

   There is a nonlinear relationship between the frame-buffer values in the image data and the intensity of the light emitted by each of the primaries on your display. This relationship is captured by a function commonly called the ``display gamma curve'' (a sample gamma curve plot is shown on the right).

   The file displayGamma.mat contains the variables:

   • gamma: is a look-up table that maps framebuffer values -> relative linear display intensity where the maximum display intensity is 1.0.
   • invGamma: is the inverse of gamma curve. It maps relative lin. disp. intensity -> frame-buffer value. This table has been interpolated to 1024 intensity levels.

   To load this file:

   load displayGamma

   ---

3. White point

   The CIELAB computation requires specifiying the white-point of the data set. In our view, each image should have its own white point. In a generally abominable practice, which we follow here, the white point is commonly set to be the white of the monitor. We don't like this much because the white point is supposed to depend on the image data, not the display device. But, nevermind.

   rgbWhite = [1 1 1]';
   whitepoint = rgb2lms* rgbWhite

Now that we have the gamma curve data and the transformation matrix rgb2lms ready, we can prepare our image data in the format S-CIELAB accepts.