I'm going to shorten variable names...
nP=size(X,3); % number planes in array X
r=zeros(nP,1); c=r; % preallocate output rows,columns arrays
for i=1:nP % iterate over planes of 3D array X
[mn,imn]=min(X(:,:,i),[],'all','linear'); % get minimum, location global minimum over plane of X
[r(i),c(i)]=ind2sub(size(X,1,2),imn); % return row,col indices of minimum in plane
end
This gives the overall minimum location for each plane; by default as your code is written you get the minimums in each column so that is size(X,2) elements but the index is then the row of each column.
That's also convertible to absolute position if that is what is wanted but it isn't clear from the post.
