10 nearest neighbors between points

Hi People,

I could really some help asap. I need to figure out how to modify the code proposed by MattJ (from a previous question) in order to find the 10 nearest points to every other point within a group of points (i.e., 10 nearest neighrbors). The points I speak of are the Fermat center points to the Triplets were found in the previous task. I have found all these "center points" and have their coordinates in the attached file (F_points.mat). Any help you could offer would be greatly appreciated.

Best

Steve

[Walter Roberson]

attachment seems to be missing?

A link to the previous Question would help.

[Steve]

Sorry about that. Here you go.

[Steve]

I would also like to plot the 10 closest points for each point with lines connecting them if possible. Thank you.

Answer 1

[Walter Roberson]

https://www.mathworks.com/help/stats/classification-using-nearest-neighbors.html

There is a specific routine for k-nearest-neighbour

[Steve]

I tried using that but could not get it to work. Is there a better way?

Attached is the intermediate file needed fo the following.

Below is the code (that I acquired from MattJ) that I'm trying to make work for this new task:

close all;
clearvars;
load('F_points.mat');
% fx = F_Points{1}(2,1);

for i = 1 : 953
    for j = 1 : i
        fy(i) = F_points{i}(1,:);
        fx(j) = F_points{j}(2,:);    
    end    
end

FermY = fy';
FermX = fx';
X = [FermX,FermY];
thresh=1000; %distance threshold
[D,K]=pdist2(X,X,'euclidean','Smallest',10);  X=X.';
 K=unique(  sort(K).',  'rows').';    K=K(:);
%  assert(numel(unique(K))==numel(K) ,'Points not well-separated enough for threshold'  );
Fs=num2cell(  reshape( X(:,K), 2,10,[] )  ,[1,2]);
Fs=Fs(:).';
C=cell2mat(cellfun(@(z)mean(z,2),Fs,'uni',0)   );
% imshow(IM); 
hold on; plot_pts2D(flipud(C)); hold off;

[Walter Roberson]

The "better way" is to use knnsearch . Your situation is suitable for using KDTreeSearcher

[Steve]

I tried it and can't get it to do what I need.

[Catalytic]

We can't see what you tried, and so can't know what to advise further.