eucilidean distance between two matrix

i have two matrix i want to find euclidean distance, for their similarity.

Answer 1

[KSSV]

doc pdist2.

Answer 2

[Walter Roberson]

It is only possible to find the euclidian distance between two matrices if the matrices are simple scalars or are vectors. If they are, for example, two dimensional histograms then euclidian distance does not apply.

[mukesh gupta]

1.

   D = sum(sum(pdist2(Z,queryica,'euclidean')));
   a(:,i)=D;

2.

% 2nd program h1 = imhist(queryica)./numel(queryica);
h2=imhist(Z)./numel(Z); 
D = sqrt(sum((h1-h2).^2));

3.

%3rd program
D=sqrt(sum(sum((queryica-Z).*(queryica-Z))));

where queryica and Z are 3X256 double matrix

Sir, which one will work as euclidian distance,

if you have other program to find euclidian distance between queryica and Z dimension 3X256.

[Walter Roberson]

Euclidian distance cannot be used in that situation.

[Walter Roberson]

What you can do is reshape() the arrays to be vectors, after which the values can act as coordinates that you can apply Euclidean distance to. But it is not clear that would have same meaning as "Euclidean distance between matrices", as the second dimension of the matrices implies a relationship between the components that is not captured by pure component-wise distance measures.

Trying to find the Euclidean distance between matrices is much like trying to sort complex numbers: a complex number is not the same as two numbers together.

Answer 3

[Image Analyst]

Not sure what your definition is, but perhaps do you mean the Euclidean distance between each corresponding pair of elements, and then maybe perhaps summing or taking of the mean of all of them?

distances = queryica - Z;
eucDistance = sum(distances); % or....
eucDistance = mean(distances(:));

If not, then maybe you can attach a diagram showing the distances between the matrices or pairs of points.