0
$\begingroup$

I have a seam of two photos joined by a seam alignment algorithm that on occasion generates duplicated pixels/objects/artifacts on both sides of the seam in the final, aligned photo. The images are large and currently the seams are checked manually for duplicate artifacts/objects. Is there an algorithm/software/programming package that will allow me to more quickly check along the seams for duplicates, preferably automatically? The algorithm should be able to spot arbitrary duplicates on both sides of the seam. What I am dealing with is sketched below as:

 |
 |
 | <---- Seam
 |
o|o <---- Duplicated object
 |
x|x <---- Duplicated object
 |
 |
Improve this question
$\endgroup$
  • $\begingroup$ wouldn't it be better to detect the duplicate objects before joining, so that your joining becomes correct? $\endgroup$ – Marcus Müller Dec 5 '19 at 21:41
  • $\begingroup$ @MarcusMüller, the seams already have some built-in overlap before they are aligned, and the alignment algorithm in use already accounts for this. I would like some sort of other program to automatically check how poor the alignment of overlap is essentially after the two images have been merged. In some cases the alignment performs perfectly and there are no visible duplicates along the seam, but in some cases the algorithm fails and we end up with these duplicate artifacts along the seam. I would like for the artifacts to be found algorithmically. $\endgroup$ – Loonuh Dec 5 '19 at 21:53
  • 1
    $\begingroup$ yeah, well, what I'd recommend would literally be "use an overlap feature detection algorithm as used in image stitching algorithms". Now, that algorithm has already failed you, so using it again wouldn't lead to new insights. Instead, tune your overlap detection to not miss these things. $\endgroup$ – Marcus Müller Dec 5 '19 at 21:56
  • $\begingroup$ I see what you're saying, apparently there are some tweaks that can be made upstream for the alignments if we know which areas are problematic though -- and right now we are finding the problem areas along the seam manually, which we would like to expedite. $\endgroup$ – Loonuh Dec 5 '19 at 22:19

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.