avir.h File Reference

The "main" inclusion file with all required classes and functions. More...

#include <cstring>
#include <cmath>
#include <stdint.h>

Go to the source code of this file.

Classes
class	avir::CBuffer< T, capint >
	Memory buffer class for element array storage, with capacity tracking. More...
class	avir::CDSPFIREQ
	FIR filter-based equalizer generator. More...
class	avir::CDSPFracFilterBankLin< fptype >
	Sinc function-based fractional delay filter bank. More...
class	avir::CDSPPeakedCosineLPF
	Low-pass filter windowed by Peaked Cosine window function. More...
class	avir::CDSPWindowGenPeakedCosine
	Peaked Cosine window function generator class. More...
class	avir::CFltBuffer
	Buffer class for parametrized low-pass filter. More...
class	avir::CImageResizer< fpclass >
	Image resizer class. More...
class	avir::CImageResizerDithererDefINL< fptype >
	Image resizer's default dithering class. More...
class	avir::CImageResizerDithererErrdINL< fptype >
	Image resizer's error-diffusion dithering class, interleaved mode. More...
class	avir::CImageResizerFilterStep< fptype, fptypeatom >
	Image resizer's filtering step class. More...
class	avir::CImageResizerFilterStepINL< fptype, fptypeatom >
	Interleaved filtering steps implementation class. More...
struct	avir::CImageResizerParams
	Resizing algorithm parameters structure. More...
struct	avir::CImageResizerParamsDef
	The default set of resizing algorithm parameters (10.06/1.88/1.029(256064.90)/0.000039). More...
struct	avir::CImageResizerParamsHigh
	Set of resizing algorithm parameters for low-aliasing resizing (11.59/1.84/1.015(73054.59)/0.000159). More...
struct	avir::CImageResizerParamsLow
	Set of resizing algorithm parameters for lower-ringing performance (9.21/1.91/1.040(391960.71)/0.000023). More...
struct	avir::CImageResizerParamsLR
	Set of resizing algorithm parameters for low-ringing performance (7.91/1.96/1.065(1980857.66)/0.000004). More...
struct	avir::CImageResizerParamsULR
	Set of resizing algorithm parameters for ultra-low-ringing performance (7.50/2.01/1.083(11568559.86)/0.000001). More...
struct	avir::CImageResizerParamsUltra
	Set of resizing algorithm parameters for ultra low-aliasing resizing (13.68/1.79/1.000(521792.07)/0.000026). More...
class	avir::CImageResizerThreadPool
	Thread pool for multi-threaded image resizing operation. More...
class	avir::CImageResizerVars
	Image resizing variables. More...
class	avir::CImageResizerVarsBase
	Image resizing variables (base class). More...
struct	avir::CImageResizerFilterStep< fptype, fptypeatom >::CResizePos
	Resizing position structure. More...
class	avir::CImageResizerFilterStep< fptype, fptypeatom >::CRPosBuf
	Resizing positions buffer class. More...
class	avir::CImageResizerFilterStep< fptype, fptypeatom >::CRPosBufArray
	Resizing positions buffer array class. More...
class	avir::CSineGen
	Sine signal generator class. More...
class	avir::CStructArray< T >
	Array of structured objects. More...
class	avir::CImageResizerThreadPool::CWorkload
	Thread pool's workload object class. More...
class	avir::fpclass_def< afptype, afptypeatom, adith >
	Floating-point processing definition and abstraction class. More...

Macros
#define	AVIR_NOCTOR(ClassName)
	Macro that defines empty copy-constructor and copy operator, with the private: prefix.
#define	AVIR_NULLPTR
	Macro is defined, if nullptr workaround is in use, for pre-C++11 compilers. Undefined at the end of file.
#define	AVIR_PI   3.1415926535897932
	The macro equals to pi constant, fills 53-bit floating point mantissa. Undefined at the end of file.
#define	AVIR_PId2   1.5707963267948966
	The macro equals to pi divided by 2 constant, fills 53-bit floating point mantissa. Undefined at the end of file.
#define	AVIR_RESIZE_PART1
#define	AVIR_RESIZE_PART1
#define	AVIR_RESIZE_PART1nx
#define	AVIR_RESIZE_PART1nx
#define	AVIR_RESIZE_PART2
#define	AVIR_RESIZE_PART2
#define	AVIR_VERSION   "3.1"
	The macro defines AVIR version string.
#define	nullptr   NULL

Functions
template<typename T1, typename T2>
void	avir::addArray (const T1 *ip, T2 *op, int l, const int ipinc=1, const int opinc=1)
	Adds values located in array ip, to array op.
template<typename T>
void	avir::calcFIRFilterResponse (const T *flt, int fltlen, const double th, double &re0, double &im0, const int fltlat=0)
	Calculates frequency response of the specified FIR filter at the specified circular frequency.
template<typename T>
T	avir::clamp (const T &Value, const T minv, const T maxv)
	"Clamps" (clips) the specified value so that it is not lesser than minv, and not greater than maxv.
template<typename T>
T	avir::convertLin2SRGB (const T s)
	Approximately de-linearizes the linear gamma value.
template<typename T, typename Tin>
T	avir::convertSRGB2Lin (const Tin &s0, const T m)
	Approximately linearizes the sRGB gamma value.
template<typename T>
T	avir::convertSRGB2Lin (const unsigned char &s0, const T)
	Approximately linearizes the sRGB gamma value. Fast specialization for uint8_t and unsigned char input.
template<typename T1, typename T2>
void	avir::copyArray (const T1 *ip, T2 *op, int l, const int ipinc=1, const int opinc=1)
	Converts (via typecast) specified array of type T1 values of length l into array of type T2 values.
template<typename T>
void	avir::normalizeFIRFilter (T *const p, const int l, const double DCGain, const int pstep=1)
	Normalizes FIR filter so that its frequency response at DC is equal to DCGain.
template<typename T>
T	avir::pow24_sRGB (const T x0)
	Power 2.4 approximation, designed for sRGB gamma correction.
template<typename T>
T	avir::pow24i_sRGB (const T x0)
	Power 1/2.4 approximation function, designed for sRGB gamma correction.
template<typename T1, typename T2>
void	avir::replicateArray (const T1 *const ip, const int ipl, T2 *op, int l, const int opinc)
	Replicates a set of adjacent elements several times in a row.
template<typename T>
T	avir::round (const T d)
	Rounding function, based on the (int) typecast. Biased result. Not suitable for numbers greater than 2^31 - 1.

Detailed Description

The "main" inclusion file with all required classes and functions.

Version

3.1

This is the "main" inclusion file for the "AVIR" image resizer. This inclusion file contains implementation of the AVIR image resizing algorithm in its entirety. Also includes several classes and functions that can be useful elsewhere.

AVIR Copyright (c) 2015-2025 Aleksey Vaneev

Macro Definition Documentation

AVIR_NOCTOR

#define AVIR_NOCTOR

(

ClassName

)

Value:

    private: \
        ClassName( const ClassName& ) { } \
        ClassName& operator = ( const ClassName& ) { return( *this ); }

Macro that defines empty copy-constructor and copy operator, with the private: prefix.

This macro should be used in classes that cannot be copied in a standard C++ way.

AVIR_RESIZE_PART1 [1/2]

#define AVIR_RESIZE_PART1

Value:

            while( rpos < rpose ) \
            { \
                const fptype x = (fptype) rpos -> x; \
                const fptype* const ftp = rpos -> ftp; \
                const fptype* const ftp2 = ftp + IntFltLen; \
                const fptype* Src = SrcLine + rpos -> SrcOffs; \
                int i;

AVIR_RESIZE_PART1 [2/2]

#define AVIR_RESIZE_PART1

Value:

            while( rpos < rpose ) \
            { \
                const fptype x = (fptype) rpos -> x; \
                const fptype* const ftp = rpos -> ftp; \
                const fptype* const ftp2 = ftp + IntFltLen0; \
                const fptype* Src = SrcLine + rpos -> SrcOffs; \
                const int IntFltLen = rpos -> fl; \
                int i;

AVIR_RESIZE_PART1nx [1/2]

#define AVIR_RESIZE_PART1nx

Value:

            while( rpos < rpose ) \
            { \
                const fptype* const ftp = rpos -> ftp; \
                const fptype* Src = SrcLine + rpos -> SrcOffs; \
                int i;

AVIR_RESIZE_PART1nx [2/2]

#define AVIR_RESIZE_PART1nx

Value:

            while( rpos < rpose ) \
            { \
                const fptype* const ftp = rpos -> ftp; \
                const fptype* Src = SrcLine + rpos -> SrcOffs; \
                const int IntFltLen = rpos -> fl; \
                int i;

AVIR_RESIZE_PART2 [1/2]

#define AVIR_RESIZE_PART2

Value:

                DstLine += DstLineIncr; \
                rpos++; \
            }

AVIR_RESIZE_PART2 [2/2]

#define AVIR_RESIZE_PART2

Value:

                DstLine += DstLineIncr; \
                rpos++; \
            }

Function Documentation

addArray()

template<typename T1, typename T2>

void avir::addArray	(	const T1 *	ip,
		T2 *	op,
		int	l,
		const int	ipinc = 1,
		const int	opinc = 1 )

Adds values located in array ip, to array op.

Parameters

	ip	Input buffer.
[out]	op	Output buffer.
	l	The number of elements to add.
	ipinc	Input buffer pointer increment.
	opinc	Output buffer pointer increment.

calcFIRFilterResponse()

template<typename T>

void avir::calcFIRFilterResponse	(	const T *	flt,
		int	fltlen,
		const double	th,
		double &	re0,
		double &	im0,
		const int	fltlat = 0 )

Calculates frequency response of the specified FIR filter at the specified circular frequency.

Phase can be calculated as atan2( im, re ). Function uses computationally-efficient oscillators instead of cos() and sin() functions.

Parameters

	flt	FIR filter's coefficients.
	fltlen	Number of coefficients (taps) in the filter.
	th	Circular frequency, [0; pi].
[out]	re0	Resulting real part of the complex frequency response.
[out]	im0	Resulting imaginary part of the complex frequency response.
	fltlat	Filter's latency in samples (taps).

Template Parameters

T	Filter coefficients' type.

clamp()

template<typename T>

T avir::clamp	(	const T &	Value,
		const T	minv,
		const T	maxv )

"Clamps" (clips) the specified value so that it is not lesser than minv, and not greater than maxv.

Parameters

Value	Value to clamp.
minv	Minimal allowed value.
maxv	Maximal allowed value.

Returns

Clamped value.

Template Parameters

T	Value's type.

convertLin2SRGB()

template<typename T>

T avir::convertLin2SRGB

(

const T

s

)

Approximately de-linearizes the linear gamma value.

Parameters

s	Linear gamma value, in the range 0 to 1.

Returns

sRGB gamma value, approximated.

Template Parameters

T	Value's type.

convertSRGB2Lin() [1/2]

template<typename T, typename Tin>

T avir::convertSRGB2Lin	(	const Tin &	s0,
		const T	m )

Approximately linearizes the sRGB gamma value.

Parameters

s0	sRGB gamma value, in the range 0 to 1.
m	Preliminary multiplier, usually a value equal to input type range.

Returns

Linearized sRGB gamma value, approximated.

Template Parameters

T	Output type.
Tin	Input value type.

convertSRGB2Lin() [2/2]

template<typename T>

T avir::convertSRGB2Lin	(	const unsigned char &	s0,
		const T	)

Approximately linearizes the sRGB gamma value. Fast specialization for uint8_t and unsigned char input.

Parameters

s0	sRGB gamma value, in the range 0 to 255.

Returns

Linearized sRGB gamma value, approximated.

Template Parameters

T	Output type.

copyArray()

template<typename T1, typename T2>

void avir::copyArray	(	const T1 *	ip,
		T2 *	op,
		int	l,
		const int	ipinc = 1,
		const int	opinc = 1 )

Converts (via typecast) specified array of type T1 values of length l into array of type T2 values.

If T1 is the same as T2, copy operation is performed. When copying data at overlapping address spaces, op should be lower than ip.

Parameters

	ip	Input buffer.
[out]	op	Output buffer.
	l	The number of elements to copy.
	ipinc	Input buffer pointer increment.
	opinc	Output buffer pointer increment.

normalizeFIRFilter()

template<typename T>

void avir::normalizeFIRFilter	(	T *const	p,
		const int	l,
		const double	DCGain,
		const int	pstep = 1 )

Normalizes FIR filter so that its frequency response at DC is equal to DCGain.

Parameters

[in,out]	p	Filter coefficients.
	l	Filter length.
	DCGain	Filter's gain at DC.
	pstep	p array step.

Template Parameters

T	Filter coefficients' type.

pow24_sRGB()

template<typename T>

T avir::pow24_sRGB

(

const T

x0

)

Power 2.4 approximation, designed for sRGB gamma correction.

Parameters

x0	Argument, in the range 0.09 to 1.

Returns

Value raised into power 2.4, approximate.

Template Parameters

T	Value's type.

pow24i_sRGB()

template<typename T>

T avir::pow24i_sRGB

(

const T

x0

)

Power 1/2.4 approximation function, designed for sRGB gamma correction.

Parameters

x0	Argument, in the range 0.003 to 1.

Returns

Value raised into power 1/2.4, approximate.

Template Parameters

T	Value's type.

replicateArray()

template<typename T1, typename T2>

void avir::replicateArray	(	const T1 *const	ip,
		const int	ipl,
		T2 *	op,
		int	l,
		const int	opinc )

Replicates a set of adjacent elements several times in a row.

This operation is usually used to replicate pixels at the start or end of image's scanline.

Parameters

	ip	Source array.
	ipl	Source array length (usually 1..4, but can be any number).
[out]	op	Destination buffer.
	l	Number of times the source array should be replicated (the destination buffer should be able to hold ipl * l number of elements).
	opinc	Destination buffer position increment after replicating the source array. This value should be equal to at least ipl.

round()

template<typename T>

T avir::round

(

const T

d

)

Rounding function, based on the (int) typecast. Biased result. Not suitable for numbers greater than 2^31 - 1.

Parameters

d	Value to round.

Returns

Rounded value. Some bias may be introduced.

Template Parameters

T	Value's type.