Retrieve original index from a masked array

Hi, I am not sure if it is even possible, but suppose I have an array A and a mask m, and I use some function (e.g. min()) to find the index of an element in the masked function, is there a way to backtrack that element back to the original unmasked array?

In other words:

[~,i] = min(A(m)); % A is an array, m is a logical array mask

Now I want to find the index j such that A(j) is the same element as (A(m))(i)

Thanks,

Answer 1

[KSSV]

Repalce the other elements in A with NaN other than mask.

B = A ; 
B(~m) = NaN ; 
[~,i] = min(B); % now the index will be of A

[Raphael Alhadef]

Perfect, thanks!

[KSSV]

Thanks is accepting/voting the answer.. 😊

[Image Analyst]

Be aware that if the min pixel value in the masked array B occurs in more than one location, i will only give you the first location it encounters, not all of them.

[Raphael Alhadef]

Yes, I also realized that this solution returns 1 if mask is all 0's which is an unintended consequence, but for my specific needs it works fine.

Thanks for the feedback!

Answer 2

[Image Analyst]

Here is a way to find out where all the min values within the masked region occur in your original masked image.

% Create a small image.
grayImage = uint8(randi(9, 20, 20))
% Create a mask that's true from rows 5-15 and columns 2-18.
mask = false(size(grayImage));
mask(5:15, 2:18) = true
% Mask the image.
maskedImage = grayImage .* uint8(mask) % Erase outside the mask.
% Get the min value of grayImage within the mask:
minValue = min(maskedImage(mask))
% Find all the rows and columns where this occurs
occurrenceMap = (maskedImage == minValue) .* mask % An image
[rows, columns] = find(occurrenceMap) % An N-by-2 list of all row,column combinations where the min occurs

[Raphael Alhadef]

Thanks! This is the original solution I thought of but it is a bit slower and I was looking for something faster.