#include <ImFusion/Base/MemImage.h>

Abstract base class for an image residing in main memory. More...

Inheritance diagram for MemImage:

Detailed Description

Abstract base class for an image residing in main memory.

Use the templated TypedImage to instantiate a MemImage for a concrete pixel type.

See also: Using Image Data, ImageProcessing

Examples: SqrtAlgorithm.cpp.

Public Types
enum class	Ownership { NotOwning = 0 , Owning = 1 }
	Enumeration to describe image data pointer ownership. More...

Public Types inherited from Image
enum	Type { BYTE = static_cast<int>(PixelType::Byte) , UBYTE = static_cast<int>(PixelType::UByte) , SHORT = static_cast<int>(PixelType::Short) , USHORT = static_cast<int>(PixelType::UShort) , INT = static_cast<int>(PixelType::Int) , UINT = static_cast<int>(PixelType::UInt) , FLOAT = static_cast<int>(PixelType::Float) , DOUBLE = static_cast<int>(PixelType::Double) , HFLOAT = static_cast<int>(PixelType::HFloat) }
	Pixel/voxel data type, equivalent to OpenGL defines. More...

enum	Location { NONE = 0 , MEMORY = 1 , OPENGL = 2 , OPENCL = 3 , CUSTOM = 6 }
	Location where the data resides. More...

Public Member Functions
std::unique_ptr< MemImage >	clone () const
	Create a deep copy of the image.

virtual bool	update (const MemImage &img)=0
	Update image data.

template<typename T>
TypedImage< T > *	typed ()
	Convenience function to perform a `dynamic_cast<TypedImage<T>>` of this instance.

template<typename T>
const TypedImage< T > *	typed () const

Location	location () const override
	Specifies that the image resides in main memory.

bool	load (const std::string &filename, int64_t offset=0)
	Loads the image content from a file, returns true if successful.

bool	save (const std::string &filename) const
	Saves the image content to a file, returns true if successful.

virtual const void *	data () const =0
	Returns a void pointer to the underlying raw pixel buffer of the image.

virtual void *	data ()=0

virtual double	valueDouble (size_t index) const =0
	Return the pixel intensity at the given index without interpolation casted to double.

virtual double	valueDouble (int x, int y, int z=0, int c=0) const =0
	Return the pixel intensity at the coordinate without interpolation casted to double.

virtual double	valueDouble (double x, double y, double z=0, int c=0, bool clamp=true, bool *valid=nullptr) const =0
	Return the pixel intensity at the given coordinate using linear interpolation casted to double.

virtual void	setValueDouble (double value, size_t index) const =0
	Set the pixel intensity at the given index to value casted to the underlying type without any clamping or normalization.

virtual void	setValueDouble (double value, int x, int y, int z=0, int c=0) const =0
	Set the pixel intensity at the given coordinate to value casted to the underlying type without any clamping or normalization.

std::pair< double, double >	getRangeDouble () const
	Computes the minimum and maximum image intensity of the entire image casted to double.

virtual void	getRangeDouble (double &minVal, double &maxVal) const =0

std::unique_ptr< MemImage >	getSliceView (int slice) const
	Extract a MemImage representing a view onto the specified slice of this image.

virtual void	fillDouble (double value)=0
	Fill image with the given double value for all channels and pixels.

virtual void	fillDouble (const std::vector< double > &channelValues)=0
	Fill image with the given double channel values for all pixels.

Public Member Functions inherited from Image
	Image (const ImageDescriptor &desc)
	Create a new Image with the given descriptor.

	Image (const Image &other)=default

Image &	operator= (const Image &other)=default

	Image (Image &&other) noexcept=default

Image &	operator= (Image &&other) noexcept=default

const ImageDescriptor &	descriptor () const
	Return the image descriptor.

void	configure (const Properties *p) override
	Configure the image descriptor from properties.

void	configuration (Properties *p) const override
	Store image descriptor into properties.

void	setShift (double val)
	Sets the intensity shift value.

void	setScale (double val)
	Sets the intensity scale value.

void	setSpacing (double sx, double sy, double sz=1.0)
	Specify pixel/voxel spacing with two or three scalar values.

void	setSpacing (double sx, double sy, double sz, bool isMetric)
	Convenience function for specifying spacing and metric flag at the same time.

void	setSpacing (const vec3 &s)
	Specify pixel/voxel spacing using a 3-vector.

void	setSpacing (const vec3 &s, bool metric)
	Convenience function for specifying spacing and metric flag at the same time.

void	setMetric (bool val)
	Set data to be metric.

double	storageToOriginal (double value) const
	Applies the image's shift and scale in order to convert value from storage pixel value domain to original pixel value domain.

double	originalToStorage (double value) const
	Applies the image's shift and scale in order to convert value from original pixel value domain to storage pixel value domain.

PixelType	pixelType () const
	Returns the data type of the image.

Type	type () const
	Returns the data type of the image.

int	width () const
	Returns the width of the image.

int	height () const
	Returns the height of the image.

int	slices () const
	Returns the number of 3D slices of the image.

int	channels () const
	Returns the number of channels of the image.

double	shift () const
	Returns the intensity shift.

double	scale () const
	Returns the intensity scale.

vec3	spacing () const
	Returns pointer to the pixel/voxel spacing values.

bool	isMetric () const
	Returns true if data is metric.

vec3	extent () const
	Returns the physical size of the image as a vector.

vec3i	dimensions () const
	Returns the dimensions of the image in pixels as a vector.

vec3	pixelToImage (const vec3 &pixel) const
	Convert a 3D pixel/voxel position to image coordinates.

vec3	imageToPixel (const vec3 &world) const
	Convert 3D image coordinates to pixel/voxel position.

mat4	pixelToImageMatrix () const
	Returns a 4x4 matrix to transform from image pixel space to image space.

mat4	imageToPixelMatrix () const
	Returns a 4x4 matrix to transform from image space to image pixel space.

size_t	index (int x, int y, int z=0, int c=0) const
	Returns a linear memory index for a pixel or voxel.

bool	hasIndex (int x, int y, int z=0, int c=0) const
	Returns true if the pixel at (x,y,z) exists, false otherwise.

vec4i	coord (size_t index) const
	Returns the pixel/voxel coordinate (x,y,z,c) for a given index.

size_t	size () const
	Returns the size (number of elements) of the image.

int	typeSize () const
	Return the nominal size in bytes of standard data types, zero if unknown.

bool	isInteger () const
	Returns whether the image is of an integer data type.

bool	isSigned () const
	Returns whether the image type supports negative values.

size_t	byteSize () const
	Returns the size of the image in bytes.

int	alignment () const
	Returns the byte alignment of an image line (up to 8)

int	dimension () const
	Returns the spatial dimension of the image.

bool	compatible (const Image &other, bool ignoreType=false, bool ignore3D=false, bool ignoreChannels=false) const
	Tells if this image is compatible to another one.

Public Member Functions inherited from Configurable
virtual void	configureDefaults ()
	Retrieve the properties of this object, replaces values with their defaults and sets it again.

void	registerParameter (ParameterBase *param)
	Register the given Parameter or SubProperty, so that it will be configured during configure()/configuration().

void	unregisterParameter (const ParameterBase *param)
	Remove the given Parameter or SubProperty from the list of registered parameters.

	Configurable (const Configurable &rhs)

	Configurable (Configurable &&rhs) noexcept

Configurable &	operator= (const Configurable &)

Configurable &	operator= (Configurable &&) noexcept

Static Public Member Functions
static std::unique_ptr< MemImage >	create (const ImageDescriptor &descriptor)
	Factory method to instantiate a MemImage from an ImageDescriptor.

static std::unique_ptr< MemImage >	create (const ImageDescriptor &descriptor, void *data, Ownership own=Ownership::NotOwning)
	Factory method to instantiate a MemImage from an ImageDescriptor and pointer to existing image data.

static std::unique_ptr< MemImage >	zeros (const ImageDescriptor &descriptor)
	Factory method to create a zero-initialized image.

Protected Member Functions
	MemImage (MemImage &&other) noexcept

MemImage &	operator= (MemImage &&other) noexcept

	MemImage (const MemImage &other)=delete

MemImage &	operator= (const MemImage &other)=delete

	MemImage (const ImageDescriptor &desc)

virtual MemImage *	cloneImpl () const =0

virtual std::unique_ptr< MemImage >	getSliceViewImpl (int slice) const =0

Protected Member Functions inherited from Image
template<typename T>
bool	typeCompatible () const

Deprecated Interface
virtual std::unique_ptr< MemImage >	prepare (bool msb=false, bool shiftOnly=false)=0
	Optimal rescaling and conversion to unsigned data type if applicable.

virtual std::unique_ptr< MemImage >	unscaled () const =0
	Return copy of this image with intensity scale and shift undone, if necessary.

Additional Inherited Members
Public Attributes inherited from Configurable
Signal	signalParametersChanged
	Emitted whenever one of the registered Parameters' or SubPropertys' `signalValueChanged` signal was emitted.

Protected Attributes inherited from Image
ImageDescriptor	m_descriptor

Protected Attributes inherited from Configurable
std::vector< Param >	m_params
	List of all registered Parameter and SubProperty instances.

Member Enumeration Documentation

◆ Ownership

enum class Ownership

strong

Enumeration to describe image data pointer ownership.

Enumerator
NotOwning	MemImage will not take ownership of the pointer, pointer must remain valid for the lifetime of the image.
Owning	MemImage will take ownership of the pointer, will delete it on destruction.

Member Function Documentation

◆ create() [1/2]

static std::unique_ptr< MemImage > create ( const ImageDescriptor & descriptor )

static

Factory method to instantiate a MemImage from an ImageDescriptor.

Note: This method does not initialize the underlying buffer Will internally dispatch to TypedImage<T>::create() based on the pixel type of descriptor.

◆ create() [2/2]

static std::unique_ptr< MemImage > create	(	const ImageDescriptor &	descriptor,
		void *	data,
		Ownership	own = Ownership::NotOwning )

static

Factory method to instantiate a MemImage from an ImageDescriptor and pointer to existing image data.

Will internally dispatch to TypedImage<T>::create() based on the pixel type of descriptor.

Parameters

descriptor	ImageDescriptor of the new image.
data	Pointer to image data, must match the descriptor layout.
own	Flag whether the newly created image shall take ownership of data.

◆ update()

virtual bool update ( const MemImage & img )

pure virtual

Update image data.

Implemented in TypedImage< T >, TypedImage< float >, TypedImage< int >, TypedImage< uint8_t >, TypedImage< unsigned char >, and TypedImage< unsigned short >.

◆ typed()

template<typename T>

TypedImage< T > * typed ( )

Convenience function to perform a dynamic_cast<TypedImage<T>> of this instance.

Returns nullptr if image is not of type T.

◆ location()

Location location ( ) const

inlineoverridevirtual

Specifies that the image resides in main memory.

Reimplemented from Image.

◆ data()

virtual const void * data ( ) const

pure virtual

Returns a void pointer to the underlying raw pixel buffer of the image.

Note: The TypedImage<T> specialization provides pointer() as type-safe alternative.

Implemented in TypedImage< T >, TypedImage< float >, TypedImage< int >, TypedImage< uint8_t >, TypedImage< unsigned char >, and TypedImage< unsigned short >.

◆ valueDouble() [1/3]

virtual double valueDouble ( size_t index ) const

pure virtual

Return the pixel intensity at the given index without interpolation casted to double.

Implemented in TypedImage< T >, TypedImage< float >, TypedImage< int >, TypedImage< uint8_t >, TypedImage< unsigned char >, and TypedImage< unsigned short >.

Examples: SqrtAlgorithm.cpp.

◆ valueDouble() [2/3]

virtual double valueDouble	(	int	x,
		int	y,
		int	z = 0,
		int	c = 0 ) const

pure virtual

Return the pixel intensity at the coordinate without interpolation casted to double.

Note: Try to avoid calling this function repeatedly because the coordinate-to-index computation is slow. When accessing entire regions of the image prefer use the index-based overload incrementing the index by a suitable stride after each function call instead of always fully recomputing the index.

Implemented in TypedImage< T >, TypedImage< float >, TypedImage< int >, TypedImage< uint8_t >, TypedImage< unsigned char >, and TypedImage< unsigned short >.

◆ valueDouble() [3/3]

virtual double valueDouble	(	double	x,
		double	y,
		double	z = 0,
		int	c = 0,
		bool	clamp = true,
		bool *	valid = nullptr ) const

pure virtual

Return the pixel intensity at the given coordinate using linear interpolation casted to double.

Note: Be aware that this function is rather slow, use it only if you really need interpolation. Consider using the set of free functions in Math::Interpolation as they provide more control on the interpolation behavior.

Implemented in TypedImage< T >, TypedImage< float >, TypedImage< int >, TypedImage< uint8_t >, TypedImage< unsigned char >, and TypedImage< unsigned short >.

◆ setValueDouble() [1/2]

virtual void setValueDouble	(	double	value,
		size_t	index ) const

pure virtual

Set the pixel intensity at the given index to value casted to the underlying type without any clamping or normalization.

Implemented in TypedImage< T >, TypedImage< float >, TypedImage< int >, TypedImage< uint8_t >, TypedImage< unsigned char >, and TypedImage< unsigned short >.

Examples: SqrtAlgorithm.cpp.

◆ setValueDouble() [2/2]

virtual void setValueDouble	(	double	value,
		int	x,
		int	y,
		int	z = 0,
		int	c = 0 ) const

pure virtual

Set the pixel intensity at the given coordinate to value casted to the underlying type without any clamping or normalization.

Note: Try to avoid calling this function repeatedly because the coordinate-to-index computation is slow. When accessing entire regions of the image prefer use the index-based overload incrementing the index by a suitable stride after each function call instead of always fully recomputing the index.

Implemented in TypedImage< T >, TypedImage< float >, TypedImage< int >, TypedImage< uint8_t >, TypedImage< unsigned char >, and TypedImage< unsigned short >.

◆ getRangeDouble()

virtual void getRangeDouble	(	double &	minVal,
		double &	maxVal ) const

pure virtual

Implemented in TypedImage< T >, TypedImage< float >, TypedImage< int >, TypedImage< uint8_t >, TypedImage< unsigned char >, and TypedImage< unsigned short >.

◆ getSliceView()

std::unique_ptr< MemImage > getSliceView ( int slice ) const

Extract a MemImage representing a view onto the specified slice of this image.

The returned image only points to the data and does not own it. It is therefore only valid as long this image is valid. Returns nullptr if slice is out of range.

◆ fillDouble() [1/2]

virtual void fillDouble ( double value )

pure virtual

Fill image with the given double value for all channels and pixels.

Note: The passed value will be casted to the underlying type of the image.

Implemented in TypedImage< T >, TypedImage< float >, TypedImage< int >, TypedImage< uint8_t >, TypedImage< unsigned char >, and TypedImage< unsigned short >.

◆ fillDouble() [2/2]

virtual void fillDouble ( const std::vector< double > & channelValues )

pure virtual

Fill image with the given double channel values for all pixels.

Note: The passed value will be casted to the underlying type of the image. The number of elements in channelValues must match the number of channels of this image.

Implemented in TypedImage< T >, TypedImage< float >, TypedImage< int >, TypedImage< uint8_t >, TypedImage< unsigned char >, and TypedImage< unsigned short >.

◆ prepare()

virtual std::unique_ptr< MemImage > prepare	(	bool	msb = false,
		bool	shiftOnly = false )

pure virtual

Optimal rescaling and conversion to unsigned data type if applicable.

Deprecated: "Use ImageProcessing::createConverted() and/or ImageProcessing::swapEndianness() instead."

Implemented in TypedImage< T >, TypedImage< float >, TypedImage< int >, TypedImage< uint8_t >, TypedImage< unsigned char >, and TypedImage< unsigned short >.

◆ unscaled()

virtual std::unique_ptr< MemImage > unscaled ( ) const

pure virtual

Return copy of this image with intensity scale and shift undone, if necessary.

The method returns a zero pointer if there is no scale and shift and if the image has a floating-point data type.

Deprecated: "Use ImageProcessing::createConverted() instead."

Implemented in TypedImage< T >, TypedImage< float >, TypedImage< int >, TypedImage< uint8_t >, TypedImage< unsigned char >, and TypedImage< unsigned short >.

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

ImFusion/Base/MemImage.h

Detailed Description

Public Types

Public Member Functions

Static Public Member Functions

Protected Member Functions

Deprecated Interface

Additional Inherited Members

Member Enumeration Documentation

◆ Ownership

Member Function Documentation

◆ create() [1/2]

◆ create() [2/2]

◆ update()

◆ typed()

◆ location()

◆ data()

◆ valueDouble() [1/3]

◆ valueDouble() [2/3]

◆ valueDouble() [3/3]

◆ setValueDouble() [1/2]

◆ setValueDouble() [2/2]

◆ getRangeDouble()

◆ getSliceView()

◆ fillDouble() [1/2]

◆ fillDouble() [2/2]

◆ prepare()

◆ unscaled()