Skip to contents

Affine pre-alignment of one object onto another

Source: R/warp-helpers.R
affine_prealign.Rd

A diffeomorphism should not have to absorb a gross change of pose and scale; that is this step's job. Both objects are centred, source is isotropically scaled to target's bounding-box diagonal, and a rigid (or affine) ICP refines the orientation. The returned apply() closure replays the same transform on any other object, so a whole set of sub-objects (e.g. neuropils) can be carried into the aligned frame consistently.

Usage

affine_prealign(
  source,
  target,
  type = c("rigid", "similarity", "affine"),
  iterations = 30L
)

Arguments

source, target

Objects accepted by nat::xyzmatrix() (mesh3d, neuron, dotprops or an N x 3 matrix). source moves onto target.

type

ICP type passed to Morpho::icpmat(): "rigid", "similarity" or "affine".

iterations

ICP iterations.

Value

A list with aligned (the transformed source), apply (a function mapping any object through the identical transform) and the components scale, centre_source, centre_target and icp (the 4x4 ICP matrix).

See also

deformetrica_register() for the non-rigid step that follows.