Euclidean distance between two structs for nearest neighbour

Hi all,

I am trying to do nearest neighbour between a set of images and a nearest neighbour model. However in my euclidean distance function I am getting errors suh as "matrix dimensions must agree".

As you can see i've attemped different ways but what im trying to do for these two functions is:

• Calculate the Euclidean distance between the test sample and all the training samples d(𝑠𝑎𝑚𝑝𝑙𝑒1,𝑠𝑎𝑚𝑝𝑙𝑒2)=|𝑠𝑎𝑚𝑝𝑙𝑒1−𝑠𝑎𝑚𝑝𝑙𝑒2|=√(𝑠𝑎𝑚𝑝𝑙𝑒1(1)−𝑠𝑎𝑚𝑝𝑙𝑒2(1))2+ (𝑠𝑎𝑚𝑝𝑙𝑒1(2)−𝑠𝑎𝑚𝑝𝑙𝑒2(2))2+⋯+(𝑠𝑎𝑚𝑝𝑙𝑒1(𝑛)−𝑠𝑎𝑚𝑝𝑙𝑒2(𝑛))2

• Select the closest training example

• Assign the closest training example’s label to the test image

function dEuc = EuclideanDistance(sample1,sample2)
%     dEuc = sqrt(sum((sample1 - sample2).^2));

%     dEuc = norm(sample1 - sample2);

%     for i = length(sample1)
%         for j = length(sample2)
%             num = sum((sample1(:) - sample2(:)).^2);
%         end
%     end
%     dEuc = sqrt(num);

%     sample1 = repmat(sample1,1,size(sample2,2));
%     dEuc = sqrt(sum((sample1(:)-sample2(:)).^2)); 
    V = sample1 - sample2;
    dEuc = sqrt(V .* V');
end

function prediction = NNTesting(testImage,modelNN)
    dataset = modelNN.neighbours;
    prediction = EuclideanDistance(testImage, dataset); 
end

Answer 1

[KSSV]

dEuc = sqrt(V .* V');

Replace the above with

dEuc = sqrt(sum(V.^2));

%% Demo

A = rand(100,2) ;  B = rand(100,2) ;

dx = A-B ;
d = sqrt(sum(dx.^2,2)) ; 

% Formula
d1 = sqrt((A(:,1)-B(:,1)).^2+(A(:,2)-B(:,2)).^2) ; 

isequal(d,d1)

You have to use sum with 1 or 2 depending on your data is row major or column major.

[Aaron Elliott]

Thanks for your reply,

When trying that I get the error "Unable to perform assignment because the indices on the left side are not compatible with the size of the right side."

[KSSV]

I have given you a demo..you need to check did you write the formula correct. We cannot help you with error unless the input dimensions and the complete code is shown along with line number of the error.

[Aaron Elliott]

Hi here are the complete files.

[Jason Reed]

You could always ask Jesus for help :)

[Aaron Elliott]

Hahaha I did, just thought to ask on here as well just have to options

[Jason Reed]

Ah, fair! Did he respond? I'm not exactly here because I figured this out haha