net.imglib2.realtransform

Class ThinplateSplineTransform

java.lang.Object

net.imglib2.realtransform.inverse.AbstractDifferentiableRealTransform

net.imglib2.realtransform.ThinplateSplineTransform

All Implemented Interfaces:

DifferentiableRealTransform, RealTransform

public class ThinplateSplineTransform
extends AbstractDifferentiableRealTransform
implements RealTransform

An n-dimensional thin plate spline transform backed by John Bogovic's jitk-tps library.

For efficiency, this implementation does not deep copies of the wrapped ThinPlateR2LogRSplineKernelTransform, since it is stateless and the same instance may be used concurrently. However, take care not to modify the wrapped instance concurrently, as that could cause race conditions.

Author:

Stephan Saalfeld, John Bogovic

Constructor Summary

Constructors

Constructor and Description
ThinplateSplineTransform(double[][] p, double[][] q)
ThinplateSplineTransform(ThinPlateR2LogRSplineKernelTransform tps)

Method Summary

Modifier and Type	Method and Description
void	apply(double[] source, double[] target) Apply the RealTransform to a source vector to obtain a target vector.
void	apply(RealLocalizable source, RealPositionable target) Apply the RealTransform to a source RealLocalizable to obtain a target RealPositionable.
ThinplateSplineTransform	copy() Create a copy of this ThinplateSplineTransform appropriate for use in concurrent code.
ThinPlateR2LogRSplineKernelTransform	getKernelTransform() Returns the instance of the wrapped ThinPlateR2LogRSplineKernelTransform.
AffineTransform	jacobian(double[] x) Returns the jacobian matrix of this transform at the point x.
int	numSourceDimensions() Returns n, the minimal number of dimension of the source vector.
int	numTargetDimensions() Returns m, the minimal dimension of the target vector.

Methods inherited from class net.imglib2.realtransform.inverse.AbstractDifferentiableRealTransform

directionToward, directionToward, matrixTranspose

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Methods inherited from interface net.imglib2.realtransform.RealTransform

apply, isIdentity

Constructor Detail

ThinplateSplineTransform

public ThinplateSplineTransform(ThinPlateR2LogRSplineKernelTransform tps)

ThinplateSplineTransform

public ThinplateSplineTransform(double[][] p,
                                double[][] q)

Method Detail

apply

public void apply(double[] source,
                  double[] target)

Description copied from interface: RealTransform

Apply the RealTransform to a source vector to obtain a target vector.

Specified by:

apply in interface RealTransform

Parameters:

source - source coordinates, length must be >= RealTransform.numSourceDimensions()

target - set this to the target coordinates, length must be >= RealTransform.numTargetDimensions()

apply

public void apply(RealLocalizable source,
                  RealPositionable target)

Description copied from interface: RealTransform

Apply the RealTransform to a source RealLocalizable to obtain a target RealPositionable.

Specified by:

apply in interface RealTransform

Parameters:

source - source coordinates, numDimensions() must be >= RealTransform.numSourceDimensions()

target - set this to the target coordinates, numDimensions() must be >= RealTransform.numTargetDimensions()

copy

public ThinplateSplineTransform copy()

Create a copy of this ThinplateSplineTransform appropriate for use in concurrent code.

For efficiency, this implementation does not make a deep copy of the wrapped ThinPlateR2LogRSplineKernelTransform, since it is stateless and the same instance may be used concurrently. However, take care not to modify the wrapped instance concurrently, as that could cause race conditions.

Specified by:

copy in interface DifferentiableRealTransform

Specified by:

copy in interface RealTransform

Returns:

a copy

numSourceDimensions

public int numSourceDimensions()

Description copied from interface: RealTransform

Returns n, the minimal number of dimension of the source vector.

Specified by:

numSourceDimensions in interface RealTransform

Returns:

the dimension of the source vector.

numTargetDimensions

public int numTargetDimensions()

Description copied from interface: RealTransform

Returns m, the minimal dimension of the target vector.

Specified by:

numTargetDimensions in interface RealTransform

Returns:

the dimension of the target vector.

jacobian

public AffineTransform jacobian(double[] x)

Description copied from class: AbstractDifferentiableRealTransform

Returns the jacobian matrix of this transform at the point x.

Specified by:

jacobian in interface DifferentiableRealTransform

Specified by:

jacobian in class AbstractDifferentiableRealTransform

Parameters:

x - the point

Returns:

the jacobian

getKernelTransform

public ThinPlateR2LogRSplineKernelTransform getKernelTransform()

Returns the instance of the wrapped ThinPlateR2LogRSplineKernelTransform.

Returns:

the kernel transform