Change Matlab colorbars' color interval manually

Question

Using three colors, and values ranging from 0-100, Colorbar is per default splitting up the colors at, respectively, 1/3 and 2/3:

I would like to select this interval manually. E.g. at 1/4 and 1/2.

MWE:

clc; close all; clear all

data = (50+50)*rand(10,3)

% green if acc(:,1)<Crit1
% red if acc(:,1)>Crit2
% yellow if acc(:,1)>Crit1 && acc(:,1)<Crit2

Crit1 = 25; 
Crit2 = 50; 

imagesc(data)
mycolormap = [0 1 0; 1 1 0; 1 0 0];
colormap(mycolormap)
colorbar

I have previously posted this question (https://math.stackexchange.com/posts/3472347/), probably in the wrong forum, without any response.

Answer 1

By default, imagesc scales the data and chooses the thresholds betwen uniformly. To change that, you have several options:

1. Define a colormap with repeated colors that produces the correspondence that you want. To get 1/4 and 1/2 you would need

   mycolormap = [0 1 0; 1 1 0; 1 0 0; 1 0 0];
   

   This gives unequal lenghts in the colorbar, according to the separation between values:

2. Apply the thresholds manually, and then change the colorbar labels:

   data = 100*rand(10,3);
   thresholds = [0 1/4 1/2 1]; % thresholds for normalized data
   mycolormap = [0 1 0; 1 1 0; 1 0 0]; % colormap with numel(threholds)-1 rows
   m = min(data(:)); % min of data
   data_normalized = data-min(data(:));
   r = max(data_normalized(:)); % range of data
   data_normalized = data_normalized./r;
   t = sum(bsxfun(@ge, data_normalized, reshape(thresholds, 1, 1, [])), 3);
   imagesc(t)
   colormap(mycolormap)
   h = colorbar;
   set(h, 'Ticks', 1:numel(thresholds))
   set(h, 'Ticklabels', m+r*thresholds)
   

   This gives equal lenghts in the colorbar, so the indicated values do not form a uniform scale. Thresholds are relative to the range of the data.

3. Same as above but with absolute thresholds:

   data = 100*rand(10,3);
   thresholds = [0 25 50 100]; % absolute thresholds for data
   mycolormap = [0 1 0; 1 1 0; 1 0 0]; % colormap with numel(threholds)-1 rows
   t = sum(bsxfun(@ge, data, reshape(thresholds, 1, 1, [])), 3);
   imagesc(t)
   colormap(mycolormap)
   h = colorbar;
   set(h, 'Ticks', 1+linspace(0, 1, numel(thresholds))*(numel(thresholds)-2))
   set(h, 'Ticklabels', thresholds)