It requires a bit of code to make this plot.
- select a suitable colormap (You might have to define your own. See colormapeditor.) Let ix=1 be white.
- map the frequences to the colormap-indicies, e.g [1:64] (should have been [2:64]). Assign 1 to missing values or whatever white means.
- use image - not imagesc
- fix the axes ticklabels
- select the colormap of your choice and add an extra row for the "missing-value-color"
- map frequency to the "colormap" and missing-value to the "extra row".
- the colorbar ticklabels will require some extra work
I use the approach described above. Pro: I get full control of the colors and the resulting code is fast. Con: requires some coding and testing.
pcolor might offer a simpler way. Try
x = [1:8];
y = [1:8];
z = x' * y ;
z(4:5,7) = nan;
pcolor( x, y, z )
Note, this gives a plot with 7x7 squares.