Differential computation on images. More...

#include <CRNDifferential.h>

Public Types
enum	RGBProjection { RGBProjection::ABSMAX, RGBProjection::SUM }

Public Member Functions
	~Differential ()=default
	Destructor. More...

	Differential (const Differential &)=delete

	Differential (Differential &&)=default

Differential &	operator= (const Differential &)=delete

Differential &	operator= (Differential &&)=default

ImageDoubleGray &	GetLx () noexcept
	Returns a reference to the internal x derivate. More...

const ImageDoubleGray &	GetLx () const
	Returns a reference to the internal x derivate. More...

ImageDoubleGray &	GetLy () noexcept
	Returns a reference to the internal y derivate. More...

const ImageDoubleGray &	GetLy () const
	Returns a reference to the internal y derivate. More...

ImageDoubleGray &	GetLx2Ly2 () noexcept
	Returns a reference to the internal first derivatives norm. More...

const ImageDoubleGray &	GetLx2Ly2 () const
	Returns a reference to the internal first derivatives norm. More...

const ImageDoubleGray &	GetLxx ()
	Returns a reference to the internal xx derivate. More...

const ImageDoubleGray &	GetLxy ()
	Returns a reference to the internal xy derivate. More...

const ImageDoubleGray &	GetLyx ()
	Returns a reference to the internal yx derivate. More...

const ImageDoubleGray &	GetLyy ()
	Returns a reference to the internal yy derivate. More...

void	Diffuse (size_t maxiter, double maxdiv=std::numeric_limits< double >::max())
	Diffuses the gradient. More...

ImageDoubleGray	MakeLw ()
	Returns the derivative of the normal to the isophotes. More...

ImageDoubleGray	MakeLvv ()
	Returns the second derivative of the tangent to the isophote. More...

ImageDoubleGray	MakeLww ()
	Returns the second derivative of the normal to the isophote. More...

ImageDoubleGray	MakeLvw ()
	Returns the cross derivative of the tangent and normal to the isophote. More...

ImageDoubleGray	MakeKappa1 ()
	Returns the first eigenvalue of the Hessian as an image. More...

ImageDoubleGray	MakeKappa2 ()
	Returns the second eigenvalue of the Hessian as an image. More...

ImageDoubleGray	MakeHessianCorner (double s=0.1)
	Returns the Hessian corner image. More...

ImageDoubleGray	MakeLaplacian ()
	Returns the Laplacian image. More...

ImageDoubleGray	MakeIsophoteCurvature ()
	Returns the isophote curvature image. More...

ImageDoubleGray	MakeFlowlineCurvature ()
	Returns the flowline curvature image. More...

ImageDoubleGray	MakeEdge ()
	Returns the edge image. More...

ImageDoubleGray	MakeCorner ()
	Returns the corner image. More...

ImageDoubleGray	MakeGaussianCurvature ()
	Returns the Gaussian curvature image. More...

ImageGradient	MakeImageGradient ()
	Returns the gradient image. More...

ImageDoubleGray	MakeGradientModule ()
	Returns the gradient module image. More...

ImageGray	MakeGradientCurvature ()
	Returns the gradient curvature image. More...

ImageDoubleGray	MakeCanny ()
	Returns Canny's edge detector image. More...

ImageDoubleGray	MakeDivergence ()
	Creates the continuous image skeleton. More...

double	AutoThreshold ()
	Computes the best threshold for lx2ly2. More...

double	GetThreshold () const noexcept
	Gets the threshold for lx2ly2. More...

void	SetThreshold (double d) noexcept
	Sets the threshold for lx2ly2. More...

bool	IsSignificant (size_t i) const noexcept
	Is a pixel significant. More...

bool	IsSignificant (size_t x, size_t y) const noexcept
	Is a pixel significant. More...

template<typename T >
void	FillNonSignificant (Image< T > &img, T val=T(0)) const
	Cleans an image of non-significant values. More...

Static Public Member Functions
static Differential	NewGaussian (const ImageRGB &src, RGBProjection proj, double sigma)
	Convolution with Gaussian derivatives. More...

static Differential	NewHalfDiff (const ImageRGB &src, RGBProjection proj)
	Alternate half derivatives. More...

static Differential	NewHalfDiffAbsMax (const ImageRGB &src, RGBProjection proj)
	AbsMax half derivatives. More...

static Differential	NewHalfDiffAbsMin (const ImageRGB &src, RGBProjection proj)
	AbsMin half derivatives. More...

template<typename T >
static Differential	NewGaussian (const Image< T > &src, double sigma)
	Convolution with Gaussian derivatives. More...

template<typename T >
static Differential	NewHalfDiff (const Image< T > &src)
	Alternate half derivatives. More...

template<typename T >
static Differential	NewHalfDiffAbsMax (const Image< T > &src)
	AbsMax half derivatives. More...

template<typename T >
static Differential	NewHalfDiffAbsMin (const Image< T > &src)
	AbsMin half derivatives. More...

static Differential	NewGaussian (const ImageDoubleGray &src, double sigma)
	Convolution with Gaussian derivatives. More...

static Differential	NewHalfDiff (const ImageDoubleGray &src)
	Alternate half derivatives. More...

static Differential	NewHalfDiffAbsMax (const ImageDoubleGray &src)
	AbsMax half derivatives. More...

static Differential	NewHalfDiffAbsMin (const ImageDoubleGray &src)
	AbsMin half derivatives. More...

Detailed Description

Differential computation on images.

This class extracts differential features from an image.

Attention: Heavy memory usage (up to 40 times the original image size)!

Author: Yann LEYDIER

Date: Jan 2015

Version: 0.5

Definition at line 50 of file CRNDifferential.h.

Member Enumeration Documentation

enum crn::Differential::RGBProjection

strong

Enumerator
ABSMAX
SUM

Definition at line 53 of file CRNDifferential.h.

Constructor & Destructor Documentation

crn::Differential::~Differential ( )

default

Destructor.

crn::Differential::Differential ( const Differential & )

delete

crn::Differential::Differential ( Differential && )

default

Member Function Documentation

double Differential::AutoThreshold ( )

Computes the best threshold for lx2ly2.

Computes the best mingrad value

Returns: The computed minimal gradient magnitude value

Todo:: find the right division factor for the deviation

Definition at line 827 of file CRNDifferential.cpp.

void Differential::Diffuse	(	size_t	maxiter,
		double	maxdiv = `std::numeric_limits<double>::max()`
	)

Diffuses the gradient.

Diffuses the gradient. All other buffers are freed in order to be recomputed if needed.

Author: Yann LEYDIER

Parameters

[in]	maxiter	Maximum number of iterations
[in]	maxdiv	Maximum divergence allowed to modify a pixel, value in ]0.0, 2.0] or >2 to always modify

Definition at line 864 of file CRNDifferential.cpp.

template<typename T >

void crn::Differential::FillNonSignificant	(	Image< T > &	img,
		T	val = `T(0)`
	)		const

inline

Cleans an image of non-significant values.

Exceptions

ExceptionDimension wrong image dimensions

Parameters

[in]	img	the image to clean
[in]	val	the value that represents non-significant pixels

Definition at line 170 of file CRNDifferential.h.

ImageDoubleGray& crn::Differential::GetLx ( )

inlinenoexcept

Returns a reference to the internal x derivate.

Definition at line 94 of file CRNDifferential.h.

const ImageDoubleGray& crn::Differential::GetLx ( ) const

inline

Returns a reference to the internal x derivate.

Definition at line 96 of file CRNDifferential.h.

ImageDoubleGray& crn::Differential::GetLx2Ly2 ( )

inlinenoexcept

Returns a reference to the internal first derivatives norm.

Definition at line 102 of file CRNDifferential.h.

const ImageDoubleGray& crn::Differential::GetLx2Ly2 ( ) const

inline

Returns a reference to the internal first derivatives norm.

Definition at line 104 of file CRNDifferential.h.

const ImageDoubleGray & Differential::GetLxx ( )

Returns a reference to the internal xx derivate.

Returns the x second derivate. Computes it if needed.

Returns: the newly created image

Definition at line 480 of file CRNDifferential.cpp.

const ImageDoubleGray & Differential::GetLxy ( )

Returns a reference to the internal xy derivate.

Returns the xy derivate. Computes it if needed.

Returns: the newly created image

Definition at line 517 of file CRNDifferential.cpp.

ImageDoubleGray& crn::Differential::GetLy ( )

inlinenoexcept

Returns a reference to the internal y derivate.

Definition at line 98 of file CRNDifferential.h.

const ImageDoubleGray& crn::Differential::GetLy ( ) const

inline

Returns a reference to the internal y derivate.

Definition at line 100 of file CRNDifferential.h.

const ImageDoubleGray & Differential::GetLyx ( )

Returns a reference to the internal yx derivate.

Returns the yx derivate. Computes it if needed.

Returns: the newly created image

Definition at line 536 of file CRNDifferential.cpp.

const ImageDoubleGray & Differential::GetLyy ( )

Returns a reference to the internal yy derivate.

Returns the y second derivate. Computes it if needed.

Returns: the newly created image

Definition at line 498 of file CRNDifferential.cpp.

double crn::Differential::GetThreshold ( ) const

inlinenoexcept

Gets the threshold for lx2ly2.

Definition at line 158 of file CRNDifferential.h.

bool crn::Differential::IsSignificant ( size_t i ) const

inlinenoexcept

Is a pixel significant.

Definition at line 162 of file CRNDifferential.h.

bool crn::Differential::IsSignificant	(	size_t	x,
		size_t	y
	)		const

inlinenoexcept

Is a pixel significant.

Definition at line 164 of file CRNDifferential.h.

ImageDoubleGray Differential::MakeCanny ( )

Returns Canny's edge detector image.

Canny's edge detector image

Returns: Canny's edge detector image

Definition at line 959 of file CRNDifferential.cpp.

ImageDoubleGray Differential::MakeCorner ( )

Returns the corner image.

Returns the corner image. Lvv * Lw²

Returns: the newly created image

Definition at line 703 of file CRNDifferential.cpp.

ImageDoubleGray Differential::MakeDivergence ( )

Creates the continuous image skeleton.

Creates a continuous skeleton from gradient divergence.

Returns: an image of the skeleton

Definition at line 911 of file CRNDifferential.cpp.

ImageDoubleGray Differential::MakeEdge ( )

Returns the edge image.

Returns the edge image. Lww / Lw

Returns: the newly created image

Definition at line 685 of file CRNDifferential.cpp.

ImageDoubleGray Differential::MakeFlowlineCurvature ( )

Returns the flowline curvature image.

Returns the flowline curvature image. Lvw / Lw

Returns: the newly created image

Definition at line 667 of file CRNDifferential.cpp.

ImageDoubleGray Differential::MakeGaussianCurvature ( )

Returns the Gaussian curvature image.

Returns the Gaussian curvature image. Lxx*Lyy - Lxy*Lyx

Returns: the newly created image

Definition at line 717 of file CRNDifferential.cpp.

ImageGray Differential::MakeGradientCurvature ( )

Returns the gradient curvature image.

Returns the gradient curvature image

Returns: the newly created image.

Definition at line 742 of file CRNDifferential.cpp.

ImageDoubleGray Differential::MakeGradientModule ( )

Returns the gradient module image.

Returns the gradient module image

Returns: the newly created image, nullptr if failed.

Definition at line 731 of file CRNDifferential.cpp.

ImageDoubleGray Differential::MakeHessianCorner ( double s = 0.1 )

Returns the Hessian corner image.

Returns the Hessian corner image. Lxx*Lyy - Lxy*Lyx - s * (Lxx+Lyy)²

Parameters

[in] s a value between 0.04 and 0.15

Returns: the newly created image

Definition at line 808 of file CRNDifferential.cpp.

ImageGradient Differential::MakeImageGradient ( )

Returns the gradient image.

Returns the gradient image

Returns: the newly created image

Definition at line 753 of file CRNDifferential.cpp.

ImageDoubleGray Differential::MakeIsophoteCurvature ( )

Returns the isophote curvature image.

Returns the isophote curvature image. Lvv / Lw

Returns: the newly created image

Definition at line 649 of file CRNDifferential.cpp.

ImageDoubleGray Differential::MakeKappa1 ( )

Returns the first eigenvalue of the Hessian as an image.

Returns the first eigenvalue of the Hessian as an image. (Lxx+Lyy + sqrt(|(Lxx+Lyy)² - 4*(Lxx*Lyy - Lxy*Lyx)|)) / 2

Returns: the newly created image

Definition at line 766 of file CRNDifferential.cpp.

ImageDoubleGray Differential::MakeKappa2 ( )

Returns the second eigenvalue of the Hessian as an image.

Returns the second eigenvalue of the Hessian as an image. (Lxx+Lyy - sqrt(|(Lxx+Lyy)² - 4*(Lxx*Lyy - Lxy*Lyx)|)) / 2

Returns: the newly created image

Definition at line 787 of file CRNDifferential.cpp.

ImageDoubleGray Differential::MakeLaplacian ( )

Returns the Laplacian image.

Returns the laplacian image. Lxx + Lyy

Returns: the newly created image

Definition at line 635 of file CRNDifferential.cpp.

ImageDoubleGray Differential::MakeLvv ( )

Returns the second derivative of the tangent to the isophote.

Returns the second derivate of the tangent to the isophotes. (Lx²*Lyy + Ly²*Lxx - Lx*Ly*(Lxy + Lyx)) / (Lx² + Ly²) Computes it if needed.

Returns: the newly created image

Definition at line 568 of file CRNDifferential.cpp.

ImageDoubleGray Differential::MakeLvw ( )

Returns the cross derivative of the tangent and normal to the isophote.

Returns the cross derivative of the tangent and the normal to the isophotes. (Lx*Ly*(Lyy-Lxx) + Lxy*Lx² - Lyx*Ly²) / (Lx² + Ly²) Computes it if needed.

Returns: the newly created image

Definition at line 612 of file CRNDifferential.cpp.

ImageDoubleGray Differential::MakeLw ( )

Returns the derivative of the normal to the isophotes.

Returns the first derivate of the normal to the isophotes. sqrt(Lx²+Ly²). Computes it if needed.

Returns: the newly created image

Definition at line 554 of file CRNDifferential.cpp.

ImageDoubleGray Differential::MakeLww ( )

Returns the second derivative of the normal to the isophote.

Returns the second derivative of the normal to the isophotes. (Lx²*Lxx + Ly²*Lyy + Lx*Ly*(Lxy + Lyx)) / (Lx² + Ly²) Computes it if needed.

Returns: the newly created image

Definition at line 590 of file CRNDifferential.cpp.

Differential Differential::NewGaussian	(	const ImageRGB &	src,
		RGBProjection	proj,
		double	sigma
	)

static

Convolution with Gaussian derivatives.

Exceptions

ExceptionDomain sigma<0

Parameters

[in]	src	the image to derivate
[in]	proj	the method of projection from R, G and B derivatives to a single derivative. If proj = ABSMAX, the 2nd derivatives are not cached and will be approximated.
[in]	sigma	the standard deviation of the Gaussian (if null, the derivation convolutions are [-1 0 1] and [1 -2 1])

Returns: a new Differential object

Definition at line 212 of file CRNDifferential.cpp.

template<typename T >

static Differential crn::Differential::NewGaussian	(	const Image< T > &	src,
		double	sigma
	)

inlinestatic

Convolution with Gaussian derivatives.

Definition at line 64 of file CRNDifferential.h.

Differential Differential::NewGaussian	(	const ImageDoubleGray &	src,
		double	sigma
	)

static

Convolution with Gaussian derivatives.

Convolution with Gaussian derivatives

Exceptions

ExceptionDomain sigma<0

Parameters

[in]	src	the image to derivate
[in]	sigma	the standard deviation of the Gaussian (if null, the derivation convolutions are [-1 0 1] and [1 -2 1])

Returns: a new Differential object

Definition at line 403 of file CRNDifferential.cpp.

Differential Differential::NewHalfDiff	(	const ImageRGB &	src,
		RGBProjection	proj
	)

static

Alternate half derivatives.

Alternate half derivatives (left and top for 1st derivatives and right and bottom for 2nd derivatives)

Parameters

[in]	src	the image to derivate
[in]	proj	the method of projection from R, G and B derivatives to a single derivative

Returns: a new Differential object

Definition at line 253 of file CRNDifferential.cpp.

template<typename T >

static Differential crn::Differential::NewHalfDiff ( const Image< T > & src )

inlinestatic

Alternate half derivatives.

Definition at line 67 of file CRNDifferential.h.

Differential Differential::NewHalfDiff ( const ImageDoubleGray & src )

static

Alternate half derivatives.

Alternate half derivatives (left and top for 1st derivatives and right and bottom for 2nd derivatives)

Parameters

[in] src the image to derivate

Returns: a new Differential object

Definition at line 413 of file CRNDifferential.cpp.

Differential Differential::NewHalfDiffAbsMax	(	const ImageRGB &	src,
		RGBProjection	proj
	)

static

AbsMax half derivatives.

AbsMax half derivatives. The 2nd derivatives are not cached and will be approximated.

Parameters

[in]	src	the image to derivate
[in]	proj	the method of projection from R, G and B derivatives to a single derivative

Definition at line 299 of file CRNDifferential.cpp.

template<typename T >

static Differential crn::Differential::NewHalfDiffAbsMax ( const Image< T > & src )

inlinestatic

AbsMax half derivatives.

Definition at line 70 of file CRNDifferential.h.

Differential Differential::NewHalfDiffAbsMax ( const ImageDoubleGray & src )

static

AbsMax half derivatives.

AbsMax half derivatives. The 2nd derivatives are not cached and will be approximated.

Parameters

[in] src the image to derivate

Returns: a new Differential object

Definition at line 430 of file CRNDifferential.cpp.

Differential Differential::NewHalfDiffAbsMin	(	const ImageRGB &	src,
		RGBProjection	proj
	)

static

AbsMin half derivatives.

AbsMin half derivatives. The 2nd derivatives are not cached and will be approximated.

Parameters

[in]	src	the image to derivate
[in]	proj	the method of projection from R, G and B derivatives to a single derivative

Definition at line 350 of file CRNDifferential.cpp.

template<typename T >

static Differential crn::Differential::NewHalfDiffAbsMin ( const Image< T > & src )

inlinestatic

AbsMin half derivatives.

Definition at line 73 of file CRNDifferential.h.

Differential Differential::NewHalfDiffAbsMin ( const ImageDoubleGray & src )

static

AbsMin half derivatives.

AbsMin half derivatives. The 2nd derivatives are not cached and will be approximated.

Parameters

[in] src the image to derivate

Returns: a new Differential object

Definition at line 449 of file CRNDifferential.cpp.

Differential& crn::Differential::operator= ( const Differential & )

delete

Differential& crn::Differential::operator= ( Differential && )

default

void crn::Differential::SetThreshold ( double d )

inlinenoexcept

Sets the threshold for lx2ly2.

Definition at line 160 of file CRNDifferential.h.

The documentation for this class was generated from the following files:

/home/travis/build/Liris-Pleiad/libcrn/src/libcrn/CRNImage/CRNDifferential.h
/home/travis/build/Liris-Pleiad/libcrn/src/libcrn/CRNImage/CRNDifferential.cpp

Public Types

Public Member Functions

Static Public Member Functions

Detailed Description

Member Enumeration Documentation

Constructor & Destructor Documentation

Member Function Documentation