image rotation

[dav1988]

ok thanks now i understand better!
anyway so we can say that it does not exist a way to estabilish if the differences are caused by a rotation or not, right?

[snibgo]

We can look at a differences image and see if the difference can be explained by a rotation. For example, we take one image and find the difference with a small rotation of that image, 0.01 degrees:

Code: Select all

convert rkdgyr.jpg ( +clone -rotate 0.01 +repage ) -gravity Center  -compose Difference -composite -auto-level rdg_rot_diff.png

We can look at rdg_rot_diff.png and see that differences occur at high-contrast object boundaries, eg around the woman, and they are strongest when the boundary is a radial from the centre. This strongly suggests a rotation around the centre.

By contrast, the differences between your original two images are very noisy, and not stronger in any direction.