Trying to implement semi-dense 3D mapping

[Loeing]

Hi Raul,

I'm trying to replicate the method used semi-dense 3D mapping from your paper and was having a little trouble conceptualizing what to use to represent epipolar line that is parametrized as a function of the horizontal coordinates Uj. Should it just be kept as the multiplication between the fundamental matrix of the 2 keyframes and the pixel ( uj = F*pj ) or should it be something else?

Any help would be appreciated. Thanks, Josh

[Loeing]

actually uj = F_pj can't be right because there needs to be an angle between L = F_pj and uj

[nsydney]

Josh, any luck getting the semi-dense mapping working?