Hi The 2D DFT computes the DFT (as defined by the equation in the fft documentation) on the columns of a, and then it computes the DFT on the rows of that output. So
a = [2 2; 0 1];
b = fft2(a);
is equivalent to
b1 = fft(a); % compute DFT of the columns of a
b2 = fft(b1.'); % compute DFT of rows of b1
