Difference compare to the paper
pmoulon opened this issue · comments
I read the implementation & the paper.
I found a difference in the way the voxelization is done.
- In the paper they normalize the scene and then use a fixed size voxelization.
- In your implementation you suppose the scene is in some given units.
I think their normalization way is preferred since it make the process coordinates system free.
We can compute the normalization factor (a scale) and them apply it to a copy of the SfM_Data scene simply by using
const double scale_factor = ...;
openMVG::geometry::Similarity3(geometry::Pose3(Mat3::Identity(), Vec3::Zero()), scale_factor);
openMVG::sfm::ApplySimilarity(sim, sfm_data);
I will make necessary changes soon. I would add the normalization in the openMVG wrapper. I would like to keep the library non openMVG specific.
I understand.
Hi @srivathsanmurali I was off during some days.
I will come back to the code slowly in the coming weeks (normalization).
Hi,
I just implemented a basic normalization in the domset library as suggested in the paper.
I will continue to optimize and improve the library. This seems like a good starting point.
I realized I might have not scaled properly.. I need to redo this :p
No worries, I'm very busy the coming week.
After I think I will be to provide a bit of help and experiment ;-)
Hey I think the scaling works properly now. Can you check it out just in case.
Will try in 2 weeks. Next week it will be very hard for me.
Sorry for the delay, but for sure I will do the testing and share back to you the result
I think the normalization work quite well. can I close this issue?
Yes you can close it