Thanks for your paper and code. But I have a question about the way to find the cam with most reconstructed points.
It's easy to understand that the counter "nr_recon[c]" should bigger than the threshold "60", but why the middle norm "normsAll[c]" should bigger than 0.06 ?
In my view, according to the bear1 cross product R*bear2 formula( bear1 x (Rbear2)), we can get a normal perpendicular to vector "bear1" and "Rbear2", and the norm of the normal is positive correlation with the area between two vector. So I think the norm smaller, the rotation "R" will be more accurately, isn't it?
Hi, Steffen
Thanks for your paper and code. But I have a question about the way to find the cam with most reconstructed points. It's easy to understand that the counter "nr_recon[c]" should bigger than the threshold "60", but why the middle norm "normsAll[c]" should bigger than 0.06 ? In my view, according to the bear1 cross product R*bear2 formula( bear1 x (Rbear2)), we can get a normal perpendicular to vector "bear1" and "Rbear2", and the norm of the normal is positive correlation with the area between two vector. So I think the norm smaller, the rotation "R" will be more accurately, isn't it?
https://github.com/urbste/MultiCol-SLAM/blob/b68853c4f8333a973450723a0899f6bfb3f4afe0/src/cMultiInitializer.cpp#L178-L191