Perhaps use some kind of downsampling on A
. To do it right you'd have to apply a low-pass filter followed by decimation, but the low-pass filter may take very long in your case. So, even if it's subject to possible aliasing, you can try to just take a sample out of n
and plot that:
n = 10; %// choose as suits you best
imagesc(x(1:n:end), y(1:n:end), A(1:n:end,1:n:end))