Package cz.vutbr.fit.layout.cormier.impl

Class CormierSegmentation

java.lang.Object

cz.vutbr.fit.layout.cormier.impl.CormierSegmentation

public class CormierSegmentation
extends Object

Implementation of the Cormier web segmentation method.

Author:

František Maštera

See Also:

Michael Cormier (2018). Computer Vision on Web Pages: A Study of Man-Made Images. UWSpace.

Constructor Summary

Constructors

Constructor	Description
CormierSegmentation(EdgeDetector edgeDetector, LineDetector lineDetector, int minSegmentLength, float signLineProbThreshold)
CormierSegmentation(EdgeDetector edgeDetector, LineDetector lineDetector, int minSegmentLength, float signLineProbThreshold, boolean parallelEnabled)

Method Summary

Modifier and Type	Method	Description
EdgeDetector	getEdgeDetector()
LineDetector	getLineDetector()
int	getMinSegmentLength()
float	getSignLineProbThreshold()
AreaTree	run(byte[] imageData, org.eclipse.rdf4j.model.IRI iri)
AreaTree	run(Page page)
void	setEdgeDetector(EdgeDetector edgeDetector)
void	setLineDetector(LineDetector lineDetector)
void	setMinSegmentLength(int minSegmentLength)
void	setSignLineProbThreshold(float signLineProbThreshold)

Methods inherited from class java.lang.Object

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

Constructor Detail

CormierSegmentation

public CormierSegmentation(EdgeDetector edgeDetector,
                           LineDetector lineDetector,
                           int minSegmentLength,
                           float signLineProbThreshold)

Parameters:

edgeDetector - Edge detector instance used for getting probabilities of locally significant edges.

lineDetector - Line detector instance used for gettine probabilities of semantically significant lines.

minSegmentLength - Minimum length of any side of each segment.

signLineProbThreshold - Minimum probability that line is significant for it to be used for segmentation.

CormierSegmentation

public CormierSegmentation(EdgeDetector edgeDetector,
                           LineDetector lineDetector,
                           int minSegmentLength,
                           float signLineProbThreshold,
                           boolean parallelEnabled)

Parameters:

parallelEnabled - If the segment(Area, Pair, Pair) parallelization will be enabled. Disabling this will cause significant slow down on multithreaded systems.

See Also:

CormierSegmentation(EdgeDetector, LineDetector, int, float)

Method Detail

getEdgeDetector

public EdgeDetector getEdgeDetector()

Returns:

Edge detector instance used for getting probabilities of locally significant edges.

setEdgeDetector

public void setEdgeDetector(EdgeDetector edgeDetector)

Parameters:

edgeDetector - Edge detector instance used for getting probabilities of locally significant edges.

getLineDetector

public LineDetector getLineDetector()

Returns:

Line detector instance used for gettine probabilities of semantically significant lines.

setLineDetector

public void setLineDetector(LineDetector lineDetector)

Parameters:

lineDetector - Line detector instance used for gettine probabilities of semantically significant lines.

getMinSegmentLength

public int getMinSegmentLength()

Returns:

Minimum length of any side of each segment.

setMinSegmentLength

public void setMinSegmentLength(int minSegmentLength)

Parameters:

minSegmentLength - Minimum length of any side of each segment.

getSignLineProbThreshold

public float getSignLineProbThreshold()

Returns:

Minimum probability that line is significant for it to be used for segmentation.

setSignLineProbThreshold

public void setSignLineProbThreshold(float signLineProbThreshold)

Parameters:

signLineProbThreshold - Minimum probability that line is significant for it to be used for segmentation.

run

public AreaTree run(Page page)
             throws IOException

Parameters:

page - Target page which will be segmented. Make sure it's possible to get a screenshot of it using Page.getPngImage().

Returns:

X-Y tree of rectangular, non-overlapping segments, where each node's children fully cover the parent segment's area, with root node covering the whole image and leaf nodes being the most granular segments.

Throws:

IOException - If failed to process the screenshot of the given page.

run

public AreaTree run(byte[] imageData,
                    org.eclipse.rdf4j.model.IRI iri)
             throws IOException

Parameters:

imageData - Screenshot of the page which will be segmented.

iri - IRI which will be assigned to the resulting AreaTree.

Returns:

X-Y tree of rectangular, non-overlapping segments, where each node's children fully cover the parent segment's area, with root node covering the whole image and leaf nodes being the most granular segments.

Throws:

IOException - If failed to process the provided image data.