Multiple Image Matching using SURF detectors

I want to compare (features) 300 images with other 300 images using SURF detectors.

With the program below, I'm facing a problem. With the 1st folder, all images are matching with only 1 image of 2nd folder.

But I want to match all 300 images with 300 images.

Can anyone help ?

path_directory_1 = 'fall';
original_files_1 = dir([path_directory_1 '/*.png']);
path_directory_2 = 'winter';
original_files_2 = dir([path_directory_2 '/*.png']);

for i = 1:length(original_files_1)
    filename = ([path_directory_1 '/' original_files_1(i).name]);
    I1 = rgb2gray(imread(filename))
    points1 = detectSURFFeatures(I1)
    [f1,vpts1] = extractFeatures(I1, points1) 
    
    for j = 1:length(original_files_2)
        filename = ([path_directory_2 '/' original_files_2(j).name]);
        I2 = rgb2gray(imread(filename))
        points2 = detectSURFFeatures(I2)
        [f2,vpts2] = extractFeatures(I2, points2)
    end
    
indexPairs = matchFeatures(f1,f2)
matchedPoints1 = vpts1(indexPairs(:,1))
matchedPoints2 = vpts2(indexPairs(:,2))
figure; ax = axes; 
showMatchedFeatures(I1, I2, matchedPoints1, matchedPoints2,'montage','Parent',ax);
legend(ax,'matchedPoints1','matchedPoints2');       
    
end

Answer 1

[Image Analyst]

Works for me. Try this more robust code and tell me what you see in the command window:

path_directory_1 = 'fall';
% path_directory_1 = pwd;
path_directory_2 = 'winter';
% path_directory_2 = pwd;

% Alert user if either folder does not exist, then exit.
if ~isfolder(path_directory_1)
	warningMessage = sprintf('Folder %s does not exist.', path_directory_1);
	uiwait(errordlg(warningMessage));
	return;
end
if ~isfolder(path_directory_2)
	warningMessage = sprintf('No %s files found in folder %s', path_directory_2);
	uiwait(errordlg(warningMessage));
	return;
end

% Get file names.
filePattern = fullfile(path_directory_1, '/*.png');
original_files_1 = dir(filePattern);
filePattern = fullfile(path_directory_2, '/*.png');
original_files_2 = dir(filePattern);

% Alert user if either folder does not have any files in it.
if isempty(original_files_1)
	warningMessage = sprintf('No %s files found in folder %s', filePattern, path_directory_1);
	uiwait(warndlg(warningMessage));
end
if isempty(original_files_2)
	warningMessage = sprintf('No %s files found in folder %s', filePattern, path_directory_2);
	uiwait(warndlg(warningMessage));
end

for i = 1:length(original_files_1)
    filename = ([path_directory_1 '/' original_files_1(i).name]);
	fprintf('\nComparing folder 1 file "%s"\n', filename);
    
    for j = 1:length(original_files_2)
        filename = ([path_directory_2 '/' original_files_2(j).name]);
		fprintf('       to folder 2 file "%s"\n', filename);
    end
end

[Aishwarya Iyengar]

Its displaying

Comparing folder 1 file "fall/image-00001.png"

to folder 2 file "winter/image-00001.png"

to folder 2 file "winter/image-00002.png"

to folder 2 file "winter/image-00003.png"

to folder 2 file "winter/image-00004.png"

to folder 2 file "winter/image-00005.png" and so on.....

Thanks alot

[Aishwarya Iyengar]

I have one more question:

For the same I have to create precision recall curve :

for that i have ground truth data

how can I create Precision-Recall curve ???

[Image Analyst]

Sorry, I don't know the definition of that term.

[Aishwarya Iyengar]

Okie no problem

Can you tell me how to construct Similarity Matrix

[Image Analyst]

Sorry, I don't know what that is and I've never used it. What do the rows and columns of the matrix represent?

[Aishwarya Iyengar]

Similarity between 2 images that should be stored in a Matrix form. Can you help in this?

Thanks in advance !

[Image Analyst]

Sorry, no. You still have not defined it. What are you using for the similarity metric? PSNR? SSIM? The Dice Sorensen coefficient? MSE? Delta E? Something else?

And you chose not to answer my direct question "What do the rows and columns of the matrix represent?" so what can I do? Not much at this point.

[Aishwarya Iyengar]

The similarity values in the similarity matrix are computed from the matchings: S(A,B) is the similarity of images A and B based on the matched features from both images.

The Matrix rows and columns represent the matched features of images in terms of binary

[Image Analyst]

Sorry, I can't help you. S() is not a built-in function of MATLAB so presumably it's a custom function that you already have so go ahead and use it. And "the matched features of images in terms of binary" makes no sense to me so perhaps an example would help.