org.openimaj.image.feature.local.affine

Class AffineSimulationExtractor<Q extends List<T>,T extends ScaleSpacePoint,I extends Image<P,I> & SinglebandImageProcessor.Processable<Float,FImage,I>,P>

java.lang.Object

org.openimaj.image.feature.local.affine.AffineSimulationExtractor<Q,T,I,P>

Type Parameters:

Q - Type of interest point list

T - Type of interest point

I - Concrete subclass of Image

P - Pixel type

Direct Known Subclasses:

ASIFT

@Reference(type=Article,
           author={"Morel, Jean-Michel","Yu, Guoshen"},
           title="{ASIFT: A New Framework for Fully Affine Invariant Image Comparison}",
           year="2009",
           journal="SIAM J. Img. Sci.",
           publisher="Society for Industrial and Applied Mathematics")
public abstract class AffineSimulationExtractor<Q extends List<T>,T extends ScaleSpacePoint,I extends Image<P,I> & SinglebandImageProcessor.Processable<Float,FImage,I>,P>
extends Object

Base class for local feature detectors/extractors that use affine simulations in order to increase detections and improve performance with respect to affine change.

Author:

Jonathon Hare (jsh2@ecs.soton.ac.uk)

Field Summary

Fields

Modifier and Type	Field and Description
Q	allInterestPoints The list of all detected interest points
protected float	BorderFact
Map<AffineParams,Q>	mappedInterestPoints The detected interest points, grouped by simulation
protected static float	PI
List<AffineParams>	simulationOrder The list of simulation parameters in the order the simulations were performed

Constructor Summary

Constructors

Constructor and Description
AffineSimulationExtractor()

Method Summary

Modifier and Type	Method and Description
protected abstract Q	detectFeatures(I image) Detect and describe the local features in the given (transformed) image.
Q	detectFeatures(I image, AffineParams params) Detect features from a single simulation.
Q	detectFeatures(I image, float theta, float tilt) Detect features from a single simulation.
void	detectFeatures(I image, int num_of_tilts) Detect features in the given image, computing the simulations based on the given number of tilts.
protected void	filterEdgesTransformed(Q keypoints, float theta, I image, float t2)
Q	getFeatures() Get the list of all the detected features
Map<AffineParams,Q>	getKeypointsMap() get the detected interest points, grouped by simulation
protected abstract Q	newList() Construct a new list capable of holding the extracted features.

Methods inherited from class java.lang.Object

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

Field Detail

PI

protected static final float PI

See Also:

Constant Field Values

BorderFact

protected float BorderFact

allInterestPoints

public Q extends List<T> allInterestPoints

The list of all detected interest points

mappedInterestPoints

public Map<AffineParams,Q extends List<T>> mappedInterestPoints

The detected interest points, grouped by simulation

simulationOrder

public List<AffineParams> simulationOrder

The list of simulation parameters in the order the simulations were performed

Constructor Detail

AffineSimulationExtractor

public AffineSimulationExtractor()

Method Detail

detectFeatures

protected abstract Q detectFeatures(I image)

Detect and describe the local features in the given (transformed) image. The returned features should be in the coordinate system of the given image; they will be automatically filtered and transformed back to the original coordinate system.

Parameters:

image - the image in which to detect features

Returns:

the local interest features

newList

protected abstract Q newList()

Construct a new list capable of holding the extracted features.

Returns:

a new list

getFeatures

public Q getFeatures()

Get the list of all the detected features

Returns:

the list of all detected features

detectFeatures

public void detectFeatures(I image,
                           int num_of_tilts)

Detect features in the given image, computing the simulations based on the given number of tilts.

Parameters:

image - the image

num_of_tilts - the number of tilt simulations

Throws:

IllegalArgumentException - if the number of tilts < 1

detectFeatures

public Q detectFeatures(I image,
                        AffineParams params)

Detect features from a single simulation.

Parameters:

image - the image

params - the simulation parameters

Returns:

the detected features

detectFeatures

public Q detectFeatures(I image,
                        float theta,
                        float tilt)

Detect features from a single simulation.

Parameters:

image - the image

theta - the rotation

tilt - the amount of tilt

Returns:

the detected features

filterEdgesTransformed

protected void filterEdgesTransformed(Q keypoints,
                                      float theta,
                                      I image,
                                      float t2)

getKeypointsMap

public Map<AffineParams,Q> getKeypointsMap()

get the detected interest points, grouped by simulation

Returns:

The detected interest points, grouped by simulation