I was given this code

function svd-figure(filename)

%

% svd-figure(filename)

%

% read in an image and spit out svd approximations

more off;

A = imread(filename);

A = double(A);

m = size(A,1);

n = size(A,2);

h = size(A,3);

for i=1:h

ranks(i) = rank(A(:,:,i));

end

rmax=min(ranks);

for i=1:h

[U(i).data,S(i).data,V(i).data] = svd(A(:,:,i));

end

indices = [1:5];

%indices = [1:10, 20:10:rmax];

for r=indices

fprintf(1,'r = %d ',r);

for i=1:h

Amod(:,:,i)=U(i).data(:,1:r)*S(i).data(1:r,1:r)*V(i).data(:,1:r)';

Adiff(:,:,i) = U(i).data(:,r)*S(i).data(r,r)*V(i).data(:,r)';

end

Amod = uint8(Amod);

Adiff = uint8(Adiff);

rstring = sprintf('%d.jpg',r);

imwrite(Amod, rstring,'jpg');

end

more on;

and use it to import a photo, with Matlabs imread command. Use the SVD to create 8:1, 4:1, and 2:1 compressed versions of the photo. Note that this compression refers to the total memory required to store the approximation. Please report how many singular values were required for each of the compression ratios.