Combine two images together

i have two different images for the same wall , i want to combine both so the second one will be placed in the proper place over the first image as shown below. any idea which code shall i use ?,i used many all converts the images to Gray.

<<

<<

.

[Michal Dobai]

This should be the good starting point: Align multiple scenes into a single image using image registration

[Mohammed Al-Mostafa]

i am trying to use

imfuse(A,Ir); imshow(ans);

but i am getting image like attached, how i can make it colored ?

[Image Analyst]

I don't think you can. imfuse() just puts images into different color channels. You'd have to find the overlap region and compute the average there, and in non overlap regions, compute the sum.

[Mohammed Al-Mostafa]

A = imread('1.jpg');
B = imread('2.jpg');

original=rgb2gray(A)
distorted=rgb2gray(B)

ptsOriginal  = detectSURFFeatures(original);
ptsDistorted = detectSURFFeatures(distorted);
[featuresOriginal,validPtsOriginal] = ...
    extractFeatures(original,ptsOriginal);
[featuresDistorted,validPtsDistorted] = ...
    extractFeatures(distorted,ptsDistorted);

index_pairs = matchFeatures(featuresOriginal,featuresDistorted);
matchedPtsOriginal  = validPtsOriginal(index_pairs(:,1));
matchedPtsDistorted = validPtsDistorted(index_pairs(:,2));
figure; 
showMatchedFeatures(original,distorted,...
    matchedPtsOriginal,matchedPtsDistorted);
title('Matched SURF points,including outliers');

[tform,inlierPtsDistorted,inlierPtsOriginal] = ...
    estimateGeometricTransform(matchedPtsDistorted,matchedPtsOriginal,...
    'similarity');
figure; 

showMatchedFeatures(original,distorted,...
    inlierPtsOriginal,inlierPtsDistorted);
title('Matched inlier points');

outputView = imref2d(size(original));
Ir = imwarp(B,tform,'OutputView',outputView);
figure; imshow(Ir); 
title('Recovered image');
saveas(gcf,'t_img4.jpg')
imwrite(Ir,'3.png');

this is the last code i used and i got this one

so i dont know which code shall i use now

[Michal Dobai]

* this comment was changed to answer *

[Image Analyst]

Thanks Michal - this was more work than I was planning on putting into it, so he's lucky you did it. Why don't you put this down in the answers section so you can get credit for you?

[Mohammed Al-Mostafa]

first thanks for the help , sorry i did a mistake with the uploading this is image 1

and this is the final result after applying the code from Michal Dobai

Answer 1

[Michal Dobai]

This is answer to your comment .

Problem with your code is, that you display only registered image (Ir). As Image Analyst suggested , you should first combine image A with registered image Ir.

So instead of:

figure; imshow(Ir);

you should do something like this:

% Calculate mask of images 1.jpg and 2.jpg
% (for this example, let's say that 'outside' black pixels are
% background pixels)
maskA = imfill(original ~= 0, 'holes');
maskB = imfill(rgb2gray(Ir) ~= 0, 'holes');

% Mask area where these 2 pictures overlap 
overlapMask = maskA & maskB;

% ...and extend it to 3 dimensional RGB image
overlapMaskRGB = cat(3, overlapMask, overlapMask, overlapMask);

% Now, as Image Analyst suggested, sum pixels in non overlap regions
% and in overlap regions, compute the average value.
% You can, for example, sum both images, and in overlaped area divide it by 2
C = A+Ir;
C(overlapMaskRGB) = C(overlapMaskRGB)/2;
figure; 
imshow(C);

Result image is attached below. Now, as you can see, there is a little problem around the edges. (there are some dark pixels). That is because in this example we 'assume' that background of your input images is completely black (that means pixel values [0 0 0]). It doesn't have to be always true, especially for JPEG images, because of the compression artifacts.

You can use one of the image formats with lossless compression (for example png) for better results, or you can set the threshold and mask pixels darker than this threshold.

[Chippy Shaji]

i have two different hazy images for the same scence i want to fused these two images which code shall i use ?

Answer 2

[Image Analyst]

Here is another link:

https://www.mathworks.com/help/vision/examples/feature-based-panoramic-image-stitching.html

and another one:

http://www.cb.uu.se/~cris/blog/index.php/archives/76