Dear all,
I would like to compute the point of a regular grid that is nearest to a given sample. I need also to keep track of the indexes of the nearest point.
I have implemented the following code, which works fine for small grids:
N = 100; %number of points
pp1 = linspace(0, 1, N);pp2 = linspace(0, 1, N);[P1, P2] = ndgrid(pp1, pp2); %regular grid of points
querypoints = rand(2,N); %points for which I want to compute the nearest neighbour
neighbours = zeros(N, 2); %it contains the indexes of the points in [P1,P2] nearest to the querypoints
distances = zeros(N*N, 1);indexes = zeros(N*N, 2); %indexes of the matrices P1 and P2 of the nearest point
for j=1:N % I compute the distances of the j-th point from the other ones
z = 1; for k=1:N for l=1:N distances(z) = norm(querypoints(:,j)-[P1(k,l);P2(k,l)]); indexes(z,:) = [k, l]; z = z + 1; end end [~, idx] = min(distances); neighbours(j,:) = indexes(idx,:);end
Unfortunately, if the number N of points increases my code become very very slow due to the three for cycles. Any idea to make my code faster (for instance through code vectorization)?
Any help is really appreciated. Thank you in advance!
Mauro
Best Answer