npstat::ConvolutionEngine1D Class Reference

#include <ConvolutionEngine1D.hh>

Public Member Functions
	ConvolutionEngine1D (unsigned dataLen, unsigned optimization=FFTW_ESTIMATE)
template<typename Real >
void	setFilter (const Real *data, unsigned dataLen, unsigned shift=0)
void	setFilterImage (const std::complex< double > *image, unsigned imLen)
template<typename Real1 , typename Real2 >
void	convolveWithFilter (const Real1 *in, Real2 *out, unsigned dataLen, unsigned shift=0)
template<typename Real1 , typename Real2 >
void	deconvolveFilter (const Real1 *in, Real2 *out, unsigned dataLen, unsigned shift=0)
template<typename Real >
void	depositFilter (unsigned long id, const Real *data, unsigned dataLen, unsigned shift=0)
void	depositFilterImage (unsigned long id, const std::complex< double > *image, unsigned imLen)
template<typename Real1 , typename Real2 >
void	convolveWithDeposit (unsigned long id, const Real1 *in, Real2 *out, unsigned dataLen, unsigned shift=0)
template<typename Real1 , typename Real2 >
void	deconvolveDeposit (unsigned long id, const Real1 *in, Real2 *out, unsigned dataLen, unsigned shift=0)
template<typename Real2 >
void	convolveFilterAndDeposit (unsigned long id, Real2 *out, unsigned dataLen)
template<typename Real2 >
void	deconvolveFilterAndDeposit (unsigned long id, Real2 *out, unsigned dataLen)
bool	discardFilter (unsigned long id)
void	clearFilterBank ()
unsigned	dataLength () const
unsigned	imageLength () const
bool	isFilterSet () const
unsigned	filterBankSize () const

Detailed Description

Class for performing convolutions by FFT using FFTW library. Internally, transforms are calculated in double precision.

Constructor & Destructor Documentation

ConvolutionEngine1D()

npstat::ConvolutionEngine1D::ConvolutionEngine1D ( unsigned  dataLen, unsigned  optimization = FFTW_ESTIMATE  )

explicit

Constructor takes data length and FFTW optimization flag as parameters. Make sure the data length is sufficient to avoid the convolution wrap-around.

Member Function Documentation

clearFilterBank()

void npstat::ConvolutionEngine1D::clearFilterBank

(

)

Discard all filters deposited by "depositFilter" method

convolveFilterAndDeposit()

template<typename Real2 >

void npstat::ConvolutionEngine1D::convolveFilterAndDeposit	(	unsigned long	id,
		Real2 *	out,
		unsigned	dataLen
	)

Convolve the filter established by the "setFilter" method with a previously deposited filter

convolveWithDeposit()

template<typename Real1 , typename Real2 >

void npstat::ConvolutionEngine1D::convolveWithDeposit	(	unsigned long	id,
		const Real1 *	in,
		Real2 *	out,
		unsigned	dataLen,
		unsigned	shift = 0
	)

Convolve provided data with one of the filters previously established by the depositFilter method. The "id" argument is used to identify a particular filter.

The "shift" argument allows you to rotate the input data buffer so that the element with index "shift" becomes new element with index 0 (essentially, the buffer is rotated left by "shift" elements).

convolveWithFilter()

template<typename Real1 , typename Real2 >

void npstat::ConvolutionEngine1D::convolveWithFilter	(	const Real1 *	in,
		Real2 *	out,
		unsigned	dataLen,
		unsigned	shift = 0
	)

Convolve provided data with the filter previously established by setFilter.

The "shift" argument allows you to rotate the input data buffer so that the element with index "shift" becomes new element with index 0 (essentially, the buffer is rotated left by "shift" elements).

dataLength()

unsigned npstat::ConvolutionEngine1D::dataLength ( ) const

inline

Expected data length

deconvolveDeposit()

template<typename Real1 , typename Real2 >

void npstat::ConvolutionEngine1D::deconvolveDeposit	(	unsigned long	id,
		const Real1 *	in,
		Real2 *	out,
		unsigned	dataLen,
		unsigned	shift = 0
	)

Deconvolve the data with a previously deposited filter

deconvolveFilter()

template<typename Real1 , typename Real2 >

void npstat::ConvolutionEngine1D::deconvolveFilter	(	const Real1 *	in,
		Real2 *	out,
		unsigned	dataLen,
		unsigned	shift = 0
	)

Deconvolve the input using the previously established filter

deconvolveFilterAndDeposit()

template<typename Real2 >

void npstat::ConvolutionEngine1D::deconvolveFilterAndDeposit	(	unsigned long	id,
		Real2 *	out,
		unsigned	dataLen
	)

Deconvolve the filter established by the "setFilter" method with a previously deposited filter

depositFilter()

template<typename Real >

void npstat::ConvolutionEngine1D::depositFilter	(	unsigned long	id,
		const Real *	data,
		unsigned	dataLen,
		unsigned	shift = 0
	)

Provide a filter for subsequent convolutions.

The "shift" argument allows you to rotate the filter data buffer so that the element with index "shift" becomes new element with index 0 (essentially, the buffer is rotated left by "shift" elements).

"id" is the user id for the filter which can be later reused with the method "convolveWithDeposit".

depositFilterImage()

void npstat::ConvolutionEngine1D::depositFilterImage	(	unsigned long	id,
		const std::complex< double > *	image,
		unsigned	imLen
	)

Provide a filter for subsequent convolutions using its Fourier space representation

"id" is the user id for the filter which can be later reused with the method "convolveWithDeposit".

discardFilter()

bool npstat::ConvolutionEngine1D::discardFilter

(

unsigned long

id

)

Discard a filter previously deposited by "depositFilter" method. This method returns "true" if the filter with the given id was indeed discarded and "false" if such a filter was not found.

imageLength()

unsigned npstat::ConvolutionEngine1D::imageLength ( ) const

inline

Expected image length

setFilter()

template<typename Real >

void npstat::ConvolutionEngine1D::setFilter	(	const Real *	data,
		unsigned	dataLen,
		unsigned	shift = 0
	)

Provide the filter for subsequent convolutions.

The "shift" argument allows you to rotate the filter data buffer so that the element with index "shift" becomes new element with index 0 (essentially, the buffer is rotated left by "shift" elements).

setFilterImage()

void npstat::ConvolutionEngine1D::setFilterImage	(	const std::complex< double > *	image,
		unsigned	imLen
	)

Provide the filter for subsequent convolutions using its Fourier space representation

---

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

• nm/ConvolutionEngine1D.hh