Find indices of neighbors in masked region - matlab

Question

I have this piece of code:
[I, J] = find(mask == 1);
for k = 1 : numel(I)
i = I(k);
j = J(k);
neighbor_ind = [i, j - 1;
i, j + 1;
i - 1, j;
i + 1, j];
end
Now I would like to find all indices s such that [I(s), J(s)] is equal to one of the rows in neighbor_ind. The neighbor indices for which this is not possible should be ignored.
How can I achieve this?
EDIT: Here is a small example.
Suppose we have the mask
0 0 0 0
0 1 1 0
0 1 1 0
0 0 0 0
(it does not have to be rectangular)
[I, J] = find(mask == 1) will give I = [2, 3, 2, 3] and J = [2, 2, 3, 3].
Now lets set i = 2, j = 2.
There are two neighbors inside the mask, and two outside. I want to know where in I, J I need to go look to find these neighbors. In this example, the solution would be I(3), J(3) for the right neighbor, and I(2), J(2) for the bottom neighbor.

Answer 1

You can use nchoosek to generate indices of neighbors.
mask = [...
0 0 0 0
0 1 1 0
0 1 1 0
0 0 0 0];
[R , C] = find(mask==1);
n = numel(R);
idx_pix = (n:-1:1).';
idx_neighbors = nchoosek(1:n,n-1);
that results:
idx_pix =
4
3
2
1
idx_neighbors =
1 2 3
1 2 4
1 3 4
2 3 4
So in your example (n == 4) neighbors of 4th pixel are [1 2 3]
in the other words :
neighbors of `idx_pix(1)` are `idx_neighbors(1,:)` : 4 ->> 1 2 3
neighbors of `idx_pix(2)` are `idx_neighbors(2,:)` : 3 ->> 1 2 4
neighbors of `idx_pix(3)` are `idx_neighbors(3,:)` : 2 ->> 1 3 4
neighbors of `idx_pix(4)` are `idx_neighbors(4,:)` : 1 ->> 2 3 4
or
neighbors of R(idx_pix(1)) are R(idx_neighbors(1,:))
...
...
...
note: It is more efficient to use linear indices instead of rows and columns so you can use this signature: IDX = find(mask==1);