Image segmentation methods. More...

Functions
template<class D >
Image< D >	Background (const Image< D > &img)
	Segments the darker clusters of the image, supposedly belonging to the background. More...

template<class D >
Image< D >	Background (const Image< D > &img, const Image< D > &canny)
	Segments the darker clusters of the image, supposedly belonging to the background. More...

template<class D >
Image< D >	BorderPixels (const Image< D > &img, const Adjacency &adj)
	Pixels belong to the border if they have at last one adjacent pixel with distinct intensity. More...

template<class D >
Image< int >	ConnectedComponents (const Image< D > &input, const Adjacency &adj)
	Returns an image with each connected component labeled with a distinct label. Connectness is defined * according to given adjacency relation. Every non-zero intensity pixel is considered belonging to the object. Only * zere intensity pixels belong to the background. More...

Image< int >	RemoveSmallComponents (const Image< int > &label, float fraction)
	The given fraction refers to the total number of pixels in the components, not to the number of * components itself. More...

template<class D >
Image< D >	OrientedGeodesicStar (const Image< D > &image, const Vector< size_t > &obj_seeds, const Vector< size_t > &bkg_seeds, double alpha=0.0, double beta=0.5)
	Returns the label from oriented geodesic segmentation algorithm. Use positive alpha values for light object and dark background, and negative alpha values for dark object and light background. More...

template<class D >
Image< D >	HighHysteresis (const Image< D > &img, const D lower, const D higher)
	Computes Hysteresis having first threshold in higher and propagating to connected pixels in lower. More...

template<class D >
void	IntensityThreshold (Image< D > &img, const D Tmin, const D Tmax)
	Computes a threshold over img. Given a pixel p, img[ p ] = 0 if img[ p ] < Tmin or img[ p ] > Tmax, and img[ p ] = 1 otherwise. More...

template<class D >
Image< int >	MinimumSpanningForest (const Image< D > &img, const Feature< D > feature, size_t regions)
	Segmentation by minimum spanning forest algorithm. More...

template<class D >
Matrix< Image< D > >	Octants (const Image< D > &data)
	Easy way to create a matrix with the 3D image or matrix octants. Octants order: x, y, z. More...

template<class D >
Image< D >	Octants (const Matrix< Image< D > > &data)
	Easy way to merge a matrix of image or matrice octants into a single image or matrix with the 3D. Octants order: x, y, z. More...

template<class D >
D	OtsuThreshold (const Image< D > &img)
	Compute otsu threshold of img. More...

template<class D >
Image< D >	Otsu (const Image< D > &img)
	Limiarize image using the OtsuThreshold( img ) limiar. More...

template<class D >
Matrix< Image< D > >	Quadrants (const Image< D > &data)
	Easy way to create a matrix with the 2D image or matrix quadrants. More...

template<class D >
Image< D >	SurfaceBorderPixels (const Image< D > &img)
	Pixels belong to the border of a surface if they have less than 7 adjacent object pixels. More...

template<class D >
Image< int >	Watershed (Image< D > &gradient, float radius=1.1f)
	Returns the label image from Watershed segmentation algorithm using local minima as the leaking points. More...

template<class D >
Image< int >	Watershed (Image< D > &gradient, const Vector< bool > &seeds)
	Returns the label image from Watershed segmentation algorithm using seeds as the leaking points. More...

template<class D >
Image< int >	Watershed (Image< D > &gradient, const Vector< size_t > &obj_seeds, const Vector< size_t > &bkg_seeds)
	Returns the label image from Watershed segmentation algorithm using seeds as the leaking points. More...

Detailed Description

Image segmentation methods.

Function Documentation

template<class D >

Image< D > Bial::Segmentation::Background ( const Image< D > & img )

Segments the darker clusters of the image, supposedly belonging to the background.

Date: 2013/Nov/29

Parameters

img	Input image.

Returns: Binary mask of the background.

Warning: none.

template<class D >

Image< D > Bial::Segmentation::Background	(	const Image< D > &	img,
		const Image< D > &	canny
	)

Segments the darker clusters of the image, supposedly belonging to the background.

Date: 2013/Nov/29

Parameters

img	Input image.
canny	Canny edges. (Suggested hysteresis thresholds: 0.8, 0.9).

Returns: Binary mask of the background.

Warning: none.

template<class D >

Image< D > Bial::Segmentation::BorderPixels	(	const Image< D > &	img,
		const Adjacency &	adj
	)

Pixels belong to the border if they have at last one adjacent pixel with distinct intensity.

Date: 2014/Jan/07

Parameters

img	Input image.
adj	An adjacency relation.

Returns: Image with border pixels only.

Warning: none.

template<class D >

Image< int > Bial::Segmentation::ConnectedComponents	(	const Image< D > &	input,
		const Adjacency &	adj
	)

Returns an image with each connected component labeled with a distinct label. Connectness is defined * according to given adjacency relation. Every non-zero intensity pixel is considered belonging to the object. Only * zere intensity pixels belong to the background.

Date: 2016/Jun/27

Parameters

input	Input image.
adj	An adjacency relation.

Returns: Image with each connected component labeled with a distinct label.

Warning: none.

template<class D >

Image< D > Bial::Segmentation::HighHysteresis	(	const Image< D > &	img,
		const D	lower,
		const D	higher
	)

Computes Hysteresis having first threshold in higher and propagating to connected pixels in lower.

Date: 2013/Nov/26

Parameters

img	Image for thresholding.
lower	lower intensity threshold value.
higher	higher intensity threshold value.

Returns: None. lower can not be greater than higher. All valid pixels receive 1.

Warning: none.

template<class D >

void Bial::Segmentation::IntensityThreshold	(	Image< D > &	img,
		const D	Tmin,
		const D	Tmax
	)

Computes a threshold over img. Given a pixel p, img[ p ] = 0 if img[ p ] < Tmin or img[ p ] > Tmax, and img[ p ] = 1 otherwise.

Date: 2013/Sep/24

Parameters

img	Image for thresholding.
Tmin	Minimum intensity.
Tmax	Maximum intensity.

Returns: None.

Warning: Tmin must be lower or equal to Tmax.

template<class D >

Image< int > Bial::Segmentation::MinimumSpanningForest	(	const Image< D > &	img,
		const Feature< D >	feature,
		size_t	regions
	)

Segmentation by minimum spanning forest algorithm.

Date: 2013/Sep/27

Parameters

img	Input image to be segmented. Used to get image dimensions.
feature	Feature vector.
regions	Number of segmented regions.

Returns: Segmented image in the number of expected regions.

Warning: Number of regions must be greater than 0 and lower than the number of samples.

template<class D >

Matrix< Image< D > > Bial::Segmentation::Octants ( const Image< D > & data )

Easy way to create a matrix with the 3D image or matrix octants. Octants order: x, y, z.

Date: 2015/Jun/16

Parameters

data	A matrix or an image.

Returns: A matrix of matrices or images with its octants for inputs with 3 dimensions.

Warning: none.

template<class D >

Image< D > Bial::Segmentation::Octants ( const Matrix< Image< D > > & data )

Easy way to merge a matrix of image or matrice octants into a single image or matrix with the 3D. Octants order: x, y, z.

Date: 2015/Aug/05

Parameters

data	A matrix of matrices or images with octants with 3 dimensions to compose the output.

Returns: Merged octants in a single matrix or image.

Warning: none.

template<class D >

Image< D > Bial::Segmentation::OrientedGeodesicStar	(	const Image< D > &	image,
		const Vector< size_t > &	obj_seeds,
		const Vector< size_t > &	bkg_seeds,
		double	alpha = `0.0`,
		double	beta = `0.5`
	)

Returns the label from oriented geodesic segmentation algorithm. Use positive alpha values for light object and dark background, and negative alpha values for dark object and light background.

Date: 2013/Nov/05

Parameters

image	Input image.
obj_seeds	Object seed coordinate vector.
bkg_seeds	Background seed coordinate vector.
alpha	Oriented edges strength. -1.0 to 1.0. Close to -1.0 (light to dark edges). Close to 0.0 (non-oriented edges). Close to 1.0 (dark to light edges).
beta	Geodesic distance force. 0.0 (geodesic distance) to 4.0 (sum function).

Returns: Label image from oriented geodesic segmentation algorithm.

Warning: none.

template<class D >

Image< D > Bial::Segmentation::Otsu ( const Image< D > & img )

Limiarize image using the OtsuThreshold( img ) limiar.

Date: 2012/Nov/01

Parameters

img	Image for thresholding.

Returns: Limiarized image.

Warning: Support only binary limiarization. Color Images will be limiarized by color channel

template<class D >

D Bial::Segmentation::OtsuThreshold ( const Image< D > & img )

Compute otsu threshold of img.

Date: 2012/Sep/08

Parameters

img	Image for thresholding.

Returns: Otsu threshold of img.

Warning: Output value may be approximated in the case of floating point images.

template<class D >

Matrix< Image< D > > Bial::Segmentation::Quadrants ( const Image< D > & data )

Easy way to create a matrix with the 2D image or matrix quadrants.

Date: 2015/Jun/16

Parameters

data	A matrix or an image.

Returns: A matrix of matrices or images with its quadrants for inputs with 2 dimensions.

Warning: none.

Image< int > Bial::Segmentation::RemoveSmallComponents	(	const Image< int > &	label,
		float	fraction
	)

The given fraction refers to the total number of pixels in the components, not to the number of * components itself.

Date: 2016/Jun/27

Parameters

input	Input labeled component image.
fraction	The fraction of components that should be removed.

Returns: Labeled component image containing only the largest compoenents, according to the given fraction.

Warning: none.

template<class D >

Image< D > Bial::Segmentation::SurfaceBorderPixels ( const Image< D > & img )

Pixels belong to the border of a surface if they have less than 7 adjacent object pixels.

Date: 2014/Jan/07

Parameters

img	Input 3D image.

Returns: Image with surface border pixels.

Warning: Used for 3D images.

template<class D >

Image< int > Bial::Segmentation::Watershed	(	Image< D > &	gradient,
		float	radius = `1.1f`
	)

Returns the label image from Watershed segmentation algorithm using local minima as the leaking points.

Date: 2013/Oct/08

Parameters

gradient	Gradient input image.
radius	Radius for local minima detection.

Returns: Label image from Watershed segmentation algorithm.

Warning: Gradient image is changed. It returns the cost map.

template<class D >

Image< int > Bial::Segmentation::Watershed	(	Image< D > &	gradient,
		const Vector< bool > &	seeds
	)

Returns the label image from Watershed segmentation algorithm using seeds as the leaking points.

Date: 2013/Oct/08

Parameters

gradient	Gradient input image.
seeds	Seed vector.

Returns: Label image from Watershed segmentation algorithm.

Warning: Gradient image is changed. It returns the cost map.

template<class D >

Image< int > Bial::Segmentation::Watershed	(	Image< D > &	gradient,
		const Vector< size_t > &	obj_seeds,
		const Vector< size_t > &	bkg_seeds
	)

Returns the label image from Watershed segmentation algorithm using seeds as the leaking points.

Date: 2013/Oct/08

Parameters

gradient	Gradient input image.
obj_seeds,bkg_seeds	Seed vectors for object and back ground respectively.

Returns: Label image from Watershed segmentation algorithm.

Warning: Gradient image is changed. It returns the cost map.

Functions

Detailed Description

Function Documentation