nearest neighbour search for scattered data

Is there any way to find 10 nearest neighbours for each point in a scattered data set? Basically I would like to do the knn search with a number of points instead of radius.

Answer 1

[Ameer Hamza]

One method is to calculate pairwise distances using pdist(), and the using mink() to calculate the ten closest points.

x = rand(100,1); % generate 100 random points
y = rand(100,1);

dists = squareform(pdist([x y]));
dists(dists==0) = inf; % so that the point does not match with itself;
[~, I] = mink(dists, 10, 2); % 100x10 matrix

Matrix I contain 100 rows (equal to the number of points) and ten columns (each entry is the index of the closest point).