Can imnoise return the indices of locations where noise was added?

HI everyone,

is there any method in Matlab which can provide the locations of corrupted pixels which are corrupted after manual insertion of noise using imnoise or anyother method?

Answer 1

[Image Analyst]

With imnoise, no, because it adds noise to every pixel, so the answer is "every pixel". With some other method, if some pixels remain untouched, you can find the changed pixels as a binary image, which is almost always the most convenient form like this:

binaryImage = originalImage ~= noisyImage;

If, for some reason, you want that in a list of coordinates, you can do that with find, like Azzi showed you

if ndims(binaryImage) == 2
  [rows, columns] = find(binaryImage
else  
  [rows, columns, colorChannels] = find(binaryImage);
end

Though this is usually less convenient. For example if you were doing a modified median filter or something, you'd use the binaryImage only, not the coordinates. What are you going to do when you say you need "the locations" so I can tell you what form of "locations" will work best for you?

Answer 2

[Azzi Abdelmalek]

If Im1 is the original image and Im2 the disturbed one, you can find the indices by

[ii,jj]=find(Im1~=Im2)

%OR

[ii,jj,kk]=find(Im1~=Im2)