fr.ign.cogit.carto.evaluation.clutter

Class RasterClutterMethod

Object

RasterClutterMethod

public class RasterClutterMethod
extends Object

Constructor Summary

Constructors

Constructor and Description
RasterClutterMethod(TiledImage image)

Method Summary

Modifier and Type	Method and Description
int	getAndExportBravoFaridClutter(float k, int minSize, float sigma, File output) Compute the Bravo & Farid (2008) scale invariant clutter (see getBravoFaridClutter()) and exports the segmented image as a file.
int	getBravoFaridClutter(float k, int minSize, float sigma) Compute the Bravo & Farid (2008) scale invariant clutter, based on the segmentation technique from Felzenszwalb and Huttenlocher (2004).
double	getEdgeDensityClutter() According to (Rosenhotlz et al 2007), a simple edge density gives a good global evaluation of an image clutter.
Map<Integer,Map<Integer,Double>>	getGridEdgeDensityClutter(int nbColumns) Compute the edge density clutter in a grid version of the image, where the number of columns of the grid is given (the size of the cells is computed according to the image size).
TiledImage	getImage()
void	setImage(TiledImage image)

Methods inherited from class Object

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

Constructor Detail

RasterClutterMethod

public RasterClutterMethod(TiledImage image)

Method Detail

getImage

public TiledImage getImage()

setImage

public void setImage(TiledImage image)

getEdgeDensityClutter

public double getEdgeDensityClutter()

According to (Rosenhotlz et al 2007), a simple edge density gives a good global evaluation of an image clutter. The edge detection method is the default JAI one, i.e. Sobel gradient magnitude.

Returns:

getGridEdgeDensityClutter

public Map<Integer,Map<Integer,Double>> getGridEdgeDensityClutter(int nbColumns)

Compute the edge density clutter in a grid version of the image, where the number of columns of the grid is given (the size of the cells is computed according to the image size).

Parameters:

nbColumns -

Returns:

getBravoFaridClutter

public int getBravoFaridClutter(float k,
                                int minSize,
                                float sigma)

Compute the Bravo & Farid (2008) scale invariant clutter, based on the segmentation technique from Felzenszwalb and Huttenlocher (2004). Once the segmentation is computed, the clutter is the number of segmented objects.

Parameters:

k - is the segmentation threshold (default 500)

minSize - is the minimum component size enforced by post-processing (default 20)

sigma - Used to smooth the input image before segmenting it (default 0.5)

Returns:

getAndExportBravoFaridClutter

public int getAndExportBravoFaridClutter(float k,
                                         int minSize,
                                         float sigma,
                                         File output)
                                  throws IOException

Compute the Bravo & Farid (2008) scale invariant clutter (see getBravoFaridClutter()) and exports the segmented image as a file.

Parameters:

k -

minSize -

sigma -

output - the file in which the image is written.

Returns:

Throws:

IOException