NaviNIBS.util.Transforms module

NaviNIBS.util.Transforms.composeTransform(R, p=None)

Return type:

ndarray

NaviNIBS.util.Transforms.decomposeTransform(A2B)

Extract translation and rotation components from transform. Assume the transform is rigid. :rtype: tuple[ndarray, ndarray] :return: (3x3 rotation matrix, 3-elem translation vector)

NaviNIBS.util.Transforms.applyTransform(A2B, pts, doCheck=True, doStrictCheck=True)

Apply 4x4 transform(s) to a set of points :type A2B: Union[ndarray, Iterable[ndarray]] :param A2B: single 4x4 transform, or iterable of 4x4 transforms. If multiple, will apply in reversed order, so that

applyTransform([space1ToSpace2Transf, space2TransfToSpace3Transf], pts) correctly transforms from space1 to space3 as might be expected with space2TransfToSpace3Transf @ space1ToSpace2Transf @ augmentedPts

Parameters:

pts (ndarray) – Nx3 points. Or can be in shape (3,) and will return transformed points with same shape.

Param:

doCheck: whether to check that transform is valid

Param:

doStrictCheck: whether raise error if transform is not valid

Return type:

ndarray

Returns:

transformed points

NaviNIBS.util.Transforms.applyDirectionTransform(A2B, dirs, doCheck=True, doStrictCheck=True)

Return type:

ndarray

NaviNIBS.util.Transforms.concatenateTransforms(A2B)

Combine transforms in reverse order, using same ordering convention as applyTransform, such that applyTransform(concatenateTransforms([space1ToSpace2Transf, space2TransfToSpace3Transf]), pts) correctly transforms from space1 to space3 as might be expected with space2TransfToSpace3Transf @ space1ToSpace2Transf @ augmentedPts

Return type:

ndarray

NaviNIBS.util.Transforms.invertTransform(A2B)

Return type:

ndarray

NaviNIBS.util.Transforms.transformToString(A2B, precision=12)

Return type:

str

NaviNIBS.util.Transforms.stringToTransform(inputStr)

Raises ValueError if unable to convert to valid transform :type inputStr: str :param inputStr: str representation of a transform, e.g. from

transformToString(np.asarray([[1, 0, 0, 0], [0, 1, 0, 0], [0, 0, 1, 0], [0, 0, 0, 1]]))

Return type:

ndarray

Returns:

transform as 4x4 ndarray

NaviNIBS.util.Transforms.calculateRotationMatrixFromTwoVectors(vecA, vecB)

Calculate rotation such that X-axis points in direction of vecA, Y-axis points in direction of (oproj_vecA vecB) adapted from https://rock-learning.github.io/pytransform3d/_apidoc/pytransform3d.rotations.matrix_from_two_vectors.html

Parameters:

• vecA (ndarray)

• vecB (ndarray)

Return type:

ndarray

Returns:

3x3 rotation matrix

NaviNIBS.util.Transforms.calculateRotationMatrixFromVectorToVector(vecA, vecB)

Calculate rotation that would rotate vecA to vecB

Return type:

ndarray

NaviNIBS.util.Transforms.estimateAligningTransform(ptsA, ptsB, method='kabsch-svd', weights=None)

Estimate a transform that aligns one set of points onto another.

Some methods assume row-wise matching of points between sets.

For details of the kabsch-svd method, see:

• https://stackoverflow.com/questions/60877274/optimal-rotation-in-3d-with-kabsch-algorithm

• https://zpl.fi/aligning-point-patterns-with-kabsch-umeyama-algorithm/

• http://nghiaho.com/?page_id=671

Parameters:

• ptsA (ndarray) – Nx3 ndarray of points

• ptsB (ndarray) – Mx3 ndarray of points

• method (str) – method to use for estimating transform. Default is ‘kabsch-svd’’

• weights (Optional[ndarray]) –

  default None, otherwise format depends on method: if method == ‘kabsch-svd’:

  Kabsch weighted algorithm will be used. Weights should be of length equal to number of points in ptsA and ptsB.

  elif method == ‘ICP’:

  Weights should be of length 6, with values as defined by simpleicp’s rbp_observation_weights argument.

Return type:

ndarray

Returns:

A2B, 4x4 transform aligning ptsA to ptsB