How to replace certain gray level values in image to find information loss in both?

I'm finding information loss in Img1 and its replica Img2 when only a certain gray level values are allowed.

Entropy_Info_Loss=Entropy(Img1)-Entropy(img2);

where img2 must not have gray values other than eg.

only_include=[20, 30, 50].

img2(img1(:)~=only_include)=0;

However my result is all zeros image of img2. Can anyone assist me..

Answer 1

[Walter Roberson]

mask = ismember(img1, only_include);
img2 = img1;
img2(~mask) = 0;

[h612]

TQ for the answer. While mask=ismember(...) gets the correct indices, img2(~mask)=0;

doesn't replace those index values with 0.

[Walter Roberson]

ismember() does not return indices for its first output: it returns a logical result that would be true for places that match. ~ of that would be the places that do not match, and img2(~mask) = 0 would replace the non-matching places with 0. It does work.

You would need more work if your img1 is an RGB image.

[h612]

Sorry, but the img2 matrix has all zero values.

[Walter Roberson]

That could happen if none of the values exactly match the list in only_include. For example if you had converted im2double() then the array values would be in the range 0 to 1.

But if you prefer a different implementation method:

img2 = zeros(size(img1), class(img1)) ;
for K = 1 : length(only_include) 
  mask = img1 == only_include(K) ;
  img2(mask) = img(mask) ;
end

[h612]

Well this did work. Thank you. I tried using the class type only at first..but it didn't work. Using for loop did. Wondering what is wrong in the previously suggested method.