Hello Ridhima
From what I gather, you are applying image registration on the same image using feature matching method and are facing an issue with the transformed image not exactly matching the original image.
Since the method of image registration being employed here is feature matching'and then warping the image using the “imwarp” function, it is expected output to have some pixel-level discrepancy between the original and transformed image. This discrepancy can arise due to interpolation artifacts and algorithmic nuances inherent in the feature matching image registration process and the implementation of the function.
Since the images are the same, you can apply the intensity-based image registration method. Following is a code snippet to use intensity-based image registration:
grayFrame1 = imread("peppers.png"); %fixed image
grayFrame1=rgb2gray(grayFrame1);
original = grayFrame1;
grayFrame2= imread("peppers.png"); %moving image
grayFrame2=rgb2gray(grayFrame2);
subplot(1,2,1)
imshow(grayFrame1)
title('Fixed Image')
subplot(1,2,2)
imshow(grayFrame2)
title('Moving Image')
[optimizer, metric] = imregconfig('monomodal');
movingRegistered = imregister(grayFrame2,grayFrame1,'rigid',optimizer, metric);
figure
imshowpair(grayFrame1, movingRegistered,'Scaling','joint')
difference =abs(original - movingRegistered);
distance = (sum(difference(:)))
For further understanding the methods mentioned above, you can refer to the following MATLAB Documentation:
- “imregister” function: https://www.mathworks.com/help/releases/R2021a/images/ref/imregister.html
- “imregconfig” function: https://www.mathworks.com/help/releases/R2021a/images/ref/imregconfig.html
I hope you find the above explanation and suggestions useful!
