org.openimaj.image.objectdetection

Class RotationSimulationObjectDetector<IMAGE extends org.openimaj.image.Image<PIXEL,IMAGE> & org.openimaj.image.processor.SinglebandImageProcessor.Processable<Float,org.openimaj.image.FImage,IMAGE>,PIXEL,DETECTED_OBJECT>

java.lang.Object

org.openimaj.image.objectdetection.RotationSimulationObjectDetector<IMAGE,PIXEL,DETECTED_OBJECT>

Type Parameters:

IMAGE - Type of image in which the detection happens

PIXEL - Type of pixel of the image in which the detection happens

DETECTED_OBJECT - The type of object emitted by the inner detector

All Implemented Interfaces:

ObjectDetector<IMAGE,TransformedDetection<DETECTED_OBJECT>>

public class RotationSimulationObjectDetector<IMAGE extends org.openimaj.image.Image<PIXEL,IMAGE> & org.openimaj.image.processor.SinglebandImageProcessor.Processable<Float,org.openimaj.image.FImage,IMAGE>,PIXEL,DETECTED_OBJECT>
extends Object
implements ObjectDetector<IMAGE,TransformedDetection<DETECTED_OBJECT>>

An ObjectDetector that wraps another ObjectDetector and performs rotation simulations on the images it passes to the internal detector. This allows a non rotation invariant detector to be able to detect rotated objects.

This implementation allows the input image to be scaled in order to reduce computational complexity, and control over the simulated angles.

Author:

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

Constructor Summary

Constructors

Constructor and Description
RotationSimulationObjectDetector(ObjectDetector<IMAGE,DETECTED_OBJECT> detector, float[] simulationAngles, float scalefactor) Construct with the given inner detector, simulation angles and scale factor.
RotationSimulationObjectDetector(ObjectDetector<IMAGE,DETECTED_OBJECT> detector, int numRotations) Construct with the given inner detector and number of rotations.
RotationSimulationObjectDetector(ObjectDetector<IMAGE,DETECTED_OBJECT> detector, int numRotations, float scalefactor) Construct with the given inner detector, number of rotations and scale factor.

Method Summary

Modifier and Type	Method and Description
List<TransformedDetection<DETECTED_OBJECT>>	detect(IMAGE image) Detect objects in the given image and return representations of them.
ObjectDetector<IMAGE,DETECTED_OBJECT>	getInnerDetector() Get the internal detector
void	setROI(org.openimaj.math.geometry.shape.Rectangle roi) (Optional operation).

Methods inherited from class java.lang.Object

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

Constructor Detail

RotationSimulationObjectDetector

public RotationSimulationObjectDetector(ObjectDetector<IMAGE,DETECTED_OBJECT> detector,
                                        int numRotations)

Construct with the given inner detector and number of rotations. Simulations will occur every 360/numRotations degrees.

Parameters:

detector - the internal detector

numRotations - the number of rotations

RotationSimulationObjectDetector

public RotationSimulationObjectDetector(ObjectDetector<IMAGE,DETECTED_OBJECT> detector,
                                        int numRotations,
                                        float scalefactor)

Construct with the given inner detector, number of rotations and scale factor. Simulations will occur every 360/numRotations degrees.

Parameters:

detector - the internal detector

numRotations - the number of rotations

scalefactor - the amount by which to scale the input image prior to passing to the inner detector

RotationSimulationObjectDetector

public RotationSimulationObjectDetector(ObjectDetector<IMAGE,DETECTED_OBJECT> detector,
                                        float[] simulationAngles,
                                        float scalefactor)

Construct with the given inner detector, simulation angles and scale factor. Simulations will occur every 360/numRotations degrees.

Parameters:

detector - the internal detector

simulationAngles - the rotation angles to simulate

scalefactor - the amount by which to scale the input image prior to passing to the inner detector

Method Detail

detect

public List<TransformedDetection<DETECTED_OBJECT>> detect(IMAGE image)

Description copied from interface: ObjectDetector

Detect objects in the given image and return representations of them. Representations often cover things like the spatial location in the image of the detected object, however, this is implementation defined.

If no objects are detected, null or an empty list may be returned.

Specified by:

detect in interface ObjectDetector<IMAGE extends org.openimaj.image.Image<PIXEL,IMAGE> & org.openimaj.image.processor.SinglebandImageProcessor.Processable<Float,org.openimaj.image.FImage,IMAGE>,TransformedDetection<DETECTED_OBJECT>>

Parameters:

image - the image to detect the object in.

Returns:

a list of detections or null if none are found.

setROI

public void setROI(org.openimaj.math.geometry.shape.Rectangle roi)

Description copied from interface: ObjectDetector

(Optional operation).

Set the region of interest within the image to search for objects. Setting to null means to search the whole image.

Specified by:

setROI in interface ObjectDetector<IMAGE extends org.openimaj.image.Image<PIXEL,IMAGE> & org.openimaj.image.processor.SinglebandImageProcessor.Processable<Float,org.openimaj.image.FImage,IMAGE>,TransformedDetection<DETECTED_OBJECT>>

Parameters:

roi - the region of interest

getInnerDetector

public ObjectDetector<IMAGE,DETECTED_OBJECT> getInnerDetector()

Get the internal detector

Returns:

the internal detector