Module: Landmark Image Warp ()

Description:

This module deforms a 3D uniform scalar field (e.g., image data) using a number of pairs of corresponding points, which are represented by a Landmarks.

Three different transformation modes are offered: Rigid transforms the input image by applying a global translation and rotation. Bookstein uses so-called thin plate splines proposed by Bookstein. Flow uses scattered data interpolation. Bookstein mode guarantees that all landmarks will be transformed exactly to their corresponding points. This is not the case for the two others. In all three modes nearest neighbor interpolation is used for resampling.

Press the Apply button to start the computation.

Connections:

Data [required]
The Landmarks, which defines corresponding points. It must contain at least 2 sets of landmarks.

ImageData [optional]
The data set (3D uniform scalar field), which is to be transformed.

Master [optional]
If this port is connected to a uniform scalar field, the output field will have the same bounding box and resolution as the master.

Ports:

Direction

Select whether points in the first set are to be moved towards their corresponding points in the second set, or the other way round.

Export Displacement Field

Export a displacement field which is defined on the output grid (). The data type of the output is given by the input data type. Usually you will need to convert the input to float first (use Convert Image Type).

Method

Method

See description above.

Premultiply Rigid

Perform a Rigid transformation followed by the nonrigid Flow transformation.

Beta

Only available in Flow mode: The larger this value is chosen, the more local and the less smooth the resulting transformation becomes. On the other hand for beta=0, the transformation is constant. In detail the basis function used in the flow method is
f(p1,p2) = exp(- d(p1,p2) * beta).

Norm

The norm to measure the distance in the flow algorithm. Either the L1 norm is used or the L2 norm .