pypose.svdtf

class pypose.svdtf(source, target)

Computes the rigid transformation ( \(SE(3)\) ) between two sets of associated point clouds (source and target) using Singular Value Decomposition (SVD).

Parameters:

• source (torch.Tensor) – the coordinates of the source point cloud. The shape has to be (…, N, 3).

• target (torch.Tensor) – the coordinates of the target point cloud. The shape has to be (…, N, 3).

Returns:

The rigid transformation matrix in SE3Type that minimizes the mean squared error between the input point sets.

Return type:

LieTensor

Warning

The number of points N has to be the same for both point clouds.

Example

>>> import torch, pypose as pp
>>> source = torch.tensor([[0., 0., 0.],
...                     [1., 0., 0.],
...                     [0., 1., 0.]])
>>> target = torch.tensor([[1., 1., 1.],
...                     [2., 1., 1.],
...                     [1., 2., 1.]])
>>> pp.svdtf(source, target)
SE3Type LieTensor:
LieTensor([1., 1., 1., 0., 0., 0., 1.])