how to find the euclidean distance between two images

how to find the euclidean distance between two images... and how to compare query image with all the images in the folder. and if there is a statistical data like mean, mode, standard deviation(more than one value) how to collect and where to store, in such a way that it can useful for comparison.

[Biruk Fikadu Gizaw]

In the following arrangement of pixels, what’s the value of the distance between the circled two points using the three distance measurements?

Answer 1

[Jan]

• Euclidean distance between two images:

Dist = sqrt(sum((image1(:) - image2(:)) .^ 2));  % [TYPO fixed, thanks Sean]

This works if the images have the same size. If now, scale one by linear of Lanczos interpolation.

• compare query image with all the images in the folder

FileList = dir(fullfile(Folder, '*.jpg'));
Result   = cell(1, numel(FileList));
for iFile = 1:numel(FileList)
  File = fullfile(Folder, FileList(iFile).name);
  Img  = imread(File);
  ... Now compare the current image with the one to be checked
  Result{iFile} = ...
end

• is there is a statistical data like mean, mode, standard deviation(more than one value) how to collect and where to store, in such a way that it can useful for comparison.

This is the wrong direction. You have to determinem, what you are looking for. The forum cannot guess, what is useful for you. Perhaps you want to recognize some vegetables, or intergalactic gas clouds, perhaps colored cows or predict, what will be the fashion for umbrellas in the next year by scanning persons in Paris from a near earth orbit. What are you looking for?

[Sean Hendryx]

Jan Simon's code is missing a right parenthesis. Here is a working version:

distance = sqrt(sum((image1(:) - image2(:)) .^ 2));

Also, here is function that computes the pixelwise distances, returning a matrix of distances where each entry in the matrix is the distance between image1(i,j) and image2(i,j):

function distanceImage = getEucDistanceImage(image1, image2)
    % returns the matrix of Euclidean distances between two images
    distanceImage = sqrt(sum((double(image1) - double(image2)).^2, 3));
end

[Walter Roberson]

Euclidean distance for an image would be more like

distanceImage = sqrt( sum(double(image1(:)) - double(image2(:))).^2) )

I suspect for the original purpose, immse() would be a better function.

[mohammed abdul wadood]

Hi, if i have 3d image (rows, columns & pixel values), how can i calculate the euclidean distance between rows of image if i assume it as vectors, or c between columns if i assume it as vectors? thanx

[Image Analyst]

So, is it a regular RGB image? Or do you have an N by 5 2-D matrix of numbers with each row being [x, y, redValue, greenValue, blueValue]? Why do you want this anyway? What do you think it will do for you?

[Walter Roberson]

rowRGB = reshape(RGBImage, size(RGBImage,1), []);
dist_between_rows = pdist(rowRGB);

[Nur Syarah Hani Ahmad Fitri]

HI JAN, i have a file of csv that have the meanPCA and its label of the image now i wanted to do recognition with new input image(going though the same process except the meanPCA of the new image is not there in the csv file yet obviously)

then how to do recognition using eucliden distance by checking the new meanPCA with the values of images that contain in the csv file .

[Image Analyst]

Jan is no longer monitoring this forum because he had some kind of problem with lag in his browser. I'm attaching code to do PCA on an RGB image so maybe you can adapt that. How did you get the mean PCA values in the first place? You might want to start a new question/thread.

Answer 2

[KSSV]

You can compute standard statistics of an image using the mean2, std2, and corr2 functions. mean2 and std2 compute the mean and standard deviation of the elements of a matrix. corr2 computes the correlation coefficient between two matrices of the same size.

https://in.mathworks.com/help/images/image-mean-standard-deviation-and-correlation-coefficent.html

[hp]

I already have a program to compute these statistics but after calcultion i required them for comparing with the statistics(all mean std, mode etc) of query image... how to do it...how to store all staticstic values in one storage so that I can use the statistic details of individual images from database and compare those with the statistics of query image

[KSSV]

images = dir('*.jpg') ; % your extenion
N = length(images) ;   % total images 
iwant = cell(N,1) ;
for i = 1:N   % loo for each image 
    % do waht you want, let iwant be std2
     iwant{i} = std2(images(i).name)
end

[Jan]

 std2(images(i).name)

The standard deviation over the characters of the file name?

[hp]

I have below code ...

using below code , query an image... and it should return set of similar images as of query image (i am having matlab7 version)

how to do it...

function colorMoments = colorMoments(image)
% input: image to be analyzed and extract 2 first moments from each R,G,B
% output: 1x6 vector containing the 2 first color momenst from each R,G,B
% channel

[filename,pathname]=uigetfile({'*.jpg';'*.png';'*.tif';'*.bmp'},'File Selector');
image=imread([pathname,filesep,filename]);

% extract color channels
R = double(image(:, :, 1));
G = double(image(:, :, 2));
B = double(image(:, :, 3));

% compute 2 first color moments from each channel
meanR = mean( R(:) );
stdR  = std( R(:) );
meanG = mean( G(:) );
stdG  = std( G(:) );
meanB = mean( B(:) );
stdB  = std( B(:) );

% construct output vector
colorMoments = zeros(1, 6);
colorMoments(1, :) = [meanR stdR meanG stdG meanB stdB];

% clear workspace
clear('R', 'G', 'B', 'meanR', 'stdR', 'meanG', 'stdG', 'meanB', 'stdB');

end