What is the advantage of EPnP?
mroe robust to noise
-> as we use more points and work with an overdetermined case…
What two equivalent problems do we have in 3D-3D geometry?
1:
N points in both first and second coordinate system
-> compute R,T between both…
2:
N+N points in a single coordinate system
-> i.e. point set moves in single coord
-> compute R,T to “compensate” movement of points
What to sub-problems do we deal with in 3D-3D geometry?
3D-3D correspondance establishment
transformation establishment
What is the difference in 3D-3D geometry when dealing with SE(3) and Sim(3)?
SE(3) -> only rotatoin and translation; same scale
Sim(3) -> Rotation, translation and scale
What is the point cloud registration problem?
given two point clouds
and 3D-3D correspondances
=> estimate the motion
What is the input to the point cloud registration problem?
two point sets in 3D (different frames)
can be obtained by e.g. triangulation or stereo vision…
can also be virtual (i.e. EPnP)
What is the minimal case solution to the point cloud registration problem (in SO(3))?
three 3D-3D correspondances
=>DLT
What is the formal definition of our 3D-3D geometry problem?
provided two sets of points (must not necesarrily be the same amount)
goal: find the optimal translation and rotation
minimizing the sum of squared error
where xi and pi are initially unknown but sought point correspondances
What two methods do we have to solve our 3D-3D geometry problem?
non-iterative method
correct correspondences are known
R, t can be directly calculated closed form (DLT)
iterative method
correct correspondences not known
generally impossible to determine optimal R,t in one step!
How does 3D-3D geometry compare with 2D-2D and 3D-2D?
2D-2D:
motion estimation from feature correspondences
-> i.e. simlar but we have 2D feature corresp. and not 3D point correspondences
minimal solution: 5 feature correspondences
popular algos:
5-point
8-point
3D-2D:
motion estimation from 3D-2D feature correspondances
-> i.e. one set in 2D features, one set in 3D points (features)
EPNP (at least 6 point corresp.)
DLT (at least 6 point corresp.)
P3P (minimal case with 3 point corresp.)
3D-3D:
both correspondences in 3D space
minimal solution: 3 point correspondences
What are pooular algos in 3D-3D?
SE(3):
non-iterative
iterative closest point (ICT)
Sim(3):
horns method
umeyamas method
What are the steps in the non-iterative method?
compute center of mass
point set normalization
SVD
decompose into rotation and translation
-> i.e. DLT…
How do we compute the center of mass and normalize in the non-iterative method?
assume cardinality is the same!
How do we perform SVD in the non-iterative method?
compute matrix W by element wise multiplication of the corresponding points (non-iterative -> we assume we know the correspondances…)
-> assumption same cardinality…
decompose it using SVD
-> where the sigmas are the singluar values
How do we decompose the minimal solution in non-iterative?
REMEMBER!
What is the idea in iterative closest point?
iteratively align two point sets
-> converges if cprresponding points are “close enough”
What is the idea of each iteration in ICP?
treat pair of points with smallest distance as temporary 3D-3D correspondence
-> given this, we can compute trhansforamtion efficiently (non-iterative method…) with SVD
=> e.g. ICP is an iteration of non-iteration method where the correspondances are the at the current iteration closest points…
What is the pipeline of ICT?
determine correspondences based on currently smallest distance
compute R,t via SVD (non-iterative method)
apply R and t to the points of the set to be registered
compute error E(R,t)
-> t is distance of projected center of mass…
if error decreased and error > threshold
repeat these steps
otherwise: stop and output final alignment
What are some variants of ICT?
weighting the correspondences (mainly for high accuracy)
rejectin outlier pairs (mainly for higher robustness)
Zuletzt geändertvor einem Jahr