Compute inverse gamma lookup table (intensity to digital control value)
result = mkInvGammaTable(gTable,numEntries)
PURPOSE:
A (display) gamma table maps digital control values to display
intensity. Don't let anyone tell you different. The gamma table of a
display is usually measured empirically or approximated by a simple rule
I = d^g + b.
To render linear intensity images on a display, requires the inverse of
a gamma table. The inverse gamma table takes linear intensity levels as
input and returns device-dependent digital control values as output.
The table must be monotonic (or else the inverse doesn't exist). For
precision, inverse gamma tables are usually defined at a finer scale of
intensities than the number of steps in the gamma table. The parameter
numEntries specifies the number of entries in the inverse table.
In some cases, measurement noise produces non-monotonicities in a gamma
table. This can happen, in particular, at low intensities. On the
assumption that non-monotonicities are meaningless, this routine sorts
non-monotonic tables and warn the user.
gTable: Gamma table from frame-buffer to linear intensity
numEntries: Number of sample values in the inverse table
result: Inverse gamma table that maps from linear intensity to
frame-buffer value.
Example:
load('displayGamma','gamma'); % This loads a gamma table
invGamma = mkInvGammaTable(gamma,4*size(gamma,1));
Test whether the tables are proper inverses. Make some linear intensity
values.
light = rand(5,3);
digital = rgb2dac(light,invGamma);
light2 = dac2rgb(digital,gamma);
percentError = 100* (light2 - light) ./ light
If we repeat the process, we avoid the imperfections of the discrete set
of look up table levels. The inversion is then perfect.
digital2 = rgb2dac(light2,invGamma)
light3 = dac2rgb(digital2,gamma)
light3 - light2
Copyright ImagEval Consultants, LLC, 2003.