Christos Choutouridis f591127e16 A V0 implementation tested against matlab
Supports:
 - HDF5 file load and store
 - Precision timing of the process to stdout
 - logging (verbose mode) to stdout
 - Command line arguments and help
2024-11-18 00:29:04 +02:00

29 行
1017 B
Matlab

function [idx, dst] = knnsearch2(C, Q, k)
% C: Is a MxD matrix (Corpus)
% Q: Is a NxD matrix (Query)
% k: The number of nearest neighbors needded
% idx: Is the Nxk matrix with the k indexes of the C points, that are
% neighbors of the nth point of Q
% dst: Is the Nxk matrix with the k distances to the C points of the nth
% point of Q
%
% Calculate the distance matrix between C and Q
% D is an m x n matrix where each element D(i, j) is the distance
% between the i-th point in C and the j-th point in Q.
% k is the number of nearest neighbors to find.
D = dist2(C, Q);
% Find the k-nearest neighbors for each query point in Q
% [~,n] = size(D);
% for j = 1:n
% [dst(:, j), idx(:, j)] = mink(D(:, j), k);
% end
[dst, idx] = mink(D, k, 1); % mink along dimension 1 for each query point
% Transpose the output to match the knnsearch format
idx = idx'; % Make idx an n x k matrix
dst = dst'; % Make dst an n x k matrix
end