This first bit of code sets up a colormap with 2*N colors ranging from dark green to light green to light red to dark red:
N = 100;
g = [zeros(N,1), linspace(100, 255, N)', zeros(N,1)] ./ 255;
r = [linspace(255, 100, N)', zeros(N,1), zeros(N,1)] ./ 255;
c = colormap([g; r]);
The below section creates a 20x20 matrix with random values between 0 and 1 (A0). Each value in A0 is then mapped to a new value in A which corresponds to its index within the colormap. Values above the cutoff are mapped to values above N (red colors), and values below the cutoff are mapped to values below N (green colors):
A0 = rand(20,20);
cutoff = 0.1;
A = A0;
Alow = A(A0<cutoff);
Ahigh = A(A0>cutoff);
biggerRange = max([range(Alow) range(Ahigh)]);
A(A0<cutoff) = N - round(N * (Alow - min(Alow)) / biggerRange);
A(A0>cutoff) = N + round(N * (Ahigh - min(Ahigh)) / biggerRange);
image(A)
The above code uses the same scale for both regions of values (above and below the cutoff). A very dark green cell will be the same distance to the cutoff as an equally dark red cell. However, for a cutoff of 0.1, all of the green cells will be very light green because their corresponding values can't be more than 0.1 away from the cutoff. If you'd prefer to scale the green and red values separately, between their respective mins and maxes, you could do so:
A0 = rand(20,20);
cutoff = 0.1;
A = A0;
Alow = A(A0<cutoff);
Ahigh = A(A0>cutoff);
A(A0<cutoff) = round(N * (Alow - min(Alow)) / (max(Alow) - min(Alow)));
A(A0>cutoff) = N + round(N * (Ahigh - min(Ahigh)) / (max(Ahigh) - min(Ahigh)));
image(A)
Here, for a cutoff of 0.1, a very dark green cell is much closer to the cutoff than an equally dark red cell, which could be misleading.
Hopefully this is helpful!
