Question

I have two grey level images and I am looking for a “proper” way to show the difference image.

The difference image should be a color image and it should show negative, zero and positive values.

Currently I am using a color map that shows negative values as blue, positive value as yellow and zero value as grey (see for example “A color map to show differences between images” http://www.cb.uu.se/~cris/blog/index.php/archives/56).

The minimum value found in the difference image is mapped to RGB=(0, 0, 1) and the maximum value found in the difference image is mapped to RGB=(0.9, 0.9, 0), the mapping is linear.

Is there any reference in the literature (academic papers, books…) to such a map?

Answer 1

For a reference, see the (archived) page that helped push these colormaps; references are at the bottom of the page.

If you want to build such a colormap, i.e. one that is linear in saturation (as opposed to e.g. hue), you need HSL2RGB from the File Exchange, and then you can construct a colormap like this:

cLength = 64; %# length of colormap
hue = [ones(floor(cLength/2),1)*0.66; ones(ceil(cLength/2),1)*0.16];
sat = [linspace(1,0,floor(cLength/2))';linspace(0,1,ceil(cLength/2))'];
lum = ones(cLength,1)*0.50;

cmap = hsl2rgb([hue,sat,lum]);

%# apply colormap
colormap(cmap)

For your visualization, if the most negative and the most positive difference are not the same, use CAXIS to ensure that the color grey maps exactly to zero.

maxAbsDiff = max(abs(differenceImage(:)));
caxis([-maxAbsDiff,maxAbsDiff])