How can i do this?

Here is my code:

[m,n] = size(I);  

for c = 1:n
    for r = 1:m
        x = round(s*r);
        y = round(s*c);
        if x > 0 && x < row && y > 0 && y < col % inside 
            S(r,c,:) = I(x,y,:);  
        end
    end
end

Implementation works fine but the timing issues is the problem.

Thanks in advance :)

[Randy Souza]

judy, did you flag your question as inappropriate for a reason? If not, can you please delete the flag? Thanks!

Answer 1

[Matt Fig]

One thing I notice in your code. It looks, by the way you index inside the FOR loop, like your image is 3D. If so, the variable col will not contain the number of columns of the matrix. It will contain the product of the number of columns and all higher dimensions. If your image is 2D, then why are you indexing like it is 3D??

I will assume your image is 3D, and you really want the variable col to store the number of columns. This vectorization is much faster than the FOR loop, depending on the sizes involved and the scale factor. Note that I use the variable IM instead of calling it image because naming a variable the same name as a MATLAB function masks that function.

C = 1:col;
R = 1:row;
X = round(R*s);
Y = round(C*s);
idx = X>0 & X<row;
X = X(idx); 
R = R(idx);
idx = Y>0 & Y<col;
Y = Y(idx);
C = C(idx);
scaled_IM2 = zeros(size(IM),class(IM));
scaled_IM2(R,C,:) = IM(X,Y,:);

Answer 2

[Sean de Wolski]

You'll get an enormous speedup just by preallocating scaled_image so that it does not change size on every iteration.

scaled_image = zeros(size(your_image));
for c
  for r 
    etc;

Also note, don't call your variable image since this is a useful builtin function.