Provides rgb/yuv colourspace mapping functions. More...

Classes
class	triple
	A 3-tuple for holding rgb or yuv triples. More...

class	cache
	A cache for a complete set of rgb/yuv mappings, amounting to about 50MB of data. More...

class	matrix_256_inverse
	Rgb-to-yuv matrix using scaled integers where 256 is 1. More...

class	matrix_256
	Yuv-to-rgb matrix using scaled integers where 256 is 1. More...

class	matrix_1000_inverse
	Rgb-to-yuv matrix using scaled integers where 1000 is 1. More...

class	matrix_1000
	Yuv-to-rgb matrix using scaled integers where 1000 is 1. More...

class	range_y
	Defines the "digital" range for y component. More...

class	range_uv
	Defines the "digital" range for u/v components. More...

class	range_rgb
	Defines the "digital" range for r/g/b components. More...

class	fp_1000
	A simple fixed-point number class with a scale factor of 1000. More...

class	fp_256
	A simple fixed-point number class with a scale factor of 256. More...

class	converter_double
	An adaptor for double. More...

class	converter_fp
	An adaptor for fp_#. More...

Functions
template<typename Tmatrix , typename Tconverter , typename Ta , typename Ttriple >
g__constexpr Tconverter::fp_type	product (Ta a1, Ta a2, Ta a3, Ttriple tfp) g__noexcept
	Does a cross-product using fp types.

template<typename Tmatrix , typename Tconverter , typename Ttriple >
g__constexpr Tconverter::fp_type	y_imp (Ttriple rgb) g__noexcept
	Calculates y from rgb using fp types. More...

template<typename Tmatrix , typename Tconverter , typename Ttriple >
g__constexpr Tconverter::fp_type	u_imp (Ttriple rgb) g__noexcept
	Calculates u (cb) from rgb using fp types. More...

template<typename Tmatrix , typename Tconverter , typename Ttriple >
g__constexpr Tconverter::fp_type	v_imp (Ttriple rgb) g__noexcept
	Calculates v (cr) from rgb using fp types. More...

template<typename Tmatrix , typename Tconverter , typename Ttriple >
g__constexpr Tconverter::fp_type	r_imp (Ttriple yuv) g__noexcept
	Calculates r from yuv using fp types. More...

template<typename Tmatrix , typename Tconverter , typename Ttriple >
g__constexpr Tconverter::fp_type	g_imp (Ttriple yuv) g__noexcept
	Calculates g from yuv using fp types. More...

template<typename Tmatrix , typename Tconverter , typename Ttriple >
g__constexpr Tconverter::fp_type	b_imp (Ttriple yuv) g__noexcept
	Calculates b from yuv using fp types. More...

template<typename Tmatrix , typename Tconverter , typename Trange_rgb , typename Trange_y , typename Trange_uv , typename Ttriple >
Ttriple	yuv_imp (Ttriple rgb) g__noexcept
	A top-level function that calculates yuv from rgb with all the implementation options exposed as template parameters.

template<typename Tmatrix , typename Tconverter , typename Trange_rgb , typename Trange_y , typename Trange_uv , typename Ttriple >
Ttriple	rgb_imp (Ttriple yuv) g__noexcept
	A top-level function that calculates rgb from yuv with all the implementation options exposed as template parameters.

triple< unsigned char >	yuv (triple< unsigned char > rgb) g__noexcept
	A top-level function that calculates yuv from rgb with default implementation options.

triple< unsigned char >	rgb (triple< unsigned char > yuv) g__noexcept
	A top-level function that calculates rgb from yuv with default implementation options.

unsigned char	r_int (unsigned char y, unsigned char, unsigned char v) g__noexcept
	A fast conversion from yuv to r.

unsigned char	g_int (unsigned char y, unsigned char u, unsigned char v) g__noexcept
	A fast conversion from yuv to g.

unsigned char	b_int (unsigned char y, unsigned char u, unsigned char) g__noexcept
	A fast conversion from yuv to b.

triple< unsigned char >	rgb_int (triple< unsigned char > yuv) g__noexcept
	A fast conversion from yuv to rgb.

unsigned char	y_int (unsigned char r, unsigned char g, unsigned char b)
	A fast conversion from rgb to y.

unsigned char	u_int (unsigned char r, unsigned char g, unsigned char b)
	A fast conversion from rgb to u.

unsigned char	v_int (unsigned char r, unsigned char g, unsigned char b)
	A fast conversion from rgb to v.

triple< unsigned char >	yuv_int (triple< unsigned char > rgb) g__noexcept
	A fast conversion from rgb to yuv.

template<typename T >
g__constexpr T	clamp (T min, T value, T max)
	Does min/max clamping.

template<typename Trange , typename Tout , typename Tin >
g__constexpr Tout	clamp (Tin value)
	Does min/max clamping to a range.

template<typename Trange , typename Tconverter >
g__constexpr Tconverter::fp_type	map_to_range_from_unsigned (typename Tconverter::fp_type fp) g__noexcept
	Scales an unsigned fp value to another fp value in the relevant range.

template<typename Trange , typename Tconverter >
g__constexpr Tconverter::fp_type	map_to_range_from_signed (typename Tconverter::fp_type fp) g__noexcept
	Scales a signed fp value to another fp value in the relevant range.

template<typename Trange , typename Tconverter >
g__constexpr Tconverter::fp_type	map_to_range_imp (typename Tconverter::fp_type fp) g__noexcept
	Scales an fp value to another fp value in the relevant range.

template<typename Trange , typename Tconverter >
g__constexpr Trange::value_type	map_to_range (typename Tconverter::fp_type fp) g__noexcept
	Maps an "analogue" space value to a "digial" range value.

template<typename Trange , typename Tconverter >
g__constexpr Tconverter::fp_type	map_to_space_signed (typename Trange::value_type n) g__noexcept
	Maps a "digital" range value to a signed "analogue" space value.

template<typename Trange , typename Tconverter >
g__constexpr Tconverter::fp_type	map_to_space_unsigned (typename Trange::value_type n) g__noexcept
	Maps a "digital" range value to an unsigned "analogue" space value.

template<typename Trange , typename Tconverter >
g__constexpr Tconverter::fp_type	map_to_space_imp (typename Trange::value_type n) g__noexcept
	Maps a "digital" range value to an "analogue" space value.

template<typename Tconverter , typename Trange1 , typename Trange2 , typename Trange3 >
g__constexpr triple< typename Tconverter::fp_type >	map_to_space (triple< typename Trange1::value_type > t) g__noexcept
	Maps a "digital" range triple to an "analogue" space triple.

fp_1000	operator* (const fp_1000 &a, const fp_1000 &b) g__noexcept

fp_1000	operator/ (const fp_1000 &a, const fp_1000 &b) g__noexcept

fp_1000	operator+ (const fp_1000 &a, const fp_1000 &b) g__noexcept

fp_1000	operator- (const fp_1000 &a, const fp_1000 &b) g__noexcept

bool	operator< (const fp_1000 &a, const fp_1000 &b) g__noexcept

bool	operator> (const fp_1000 &a, const fp_1000 &b) g__noexcept

fp_256	operator* (fp_256 a, fp_256 b) g__noexcept

fp_256	operator/ (fp_256 a, fp_256 b) g__noexcept

fp_256	operator+ (fp_256 a, fp_256 b) g__noexcept

fp_256	operator- (fp_256 a, fp_256 b) g__noexcept

bool	operator< (fp_256 a, fp_256 b) g__noexcept

bool	operator> (fp_256 a, fp_256 b) g__noexcept

Detailed Description

Provides rgb/yuv colourspace mapping functions.

Synopsis.

typedef Gr::ColourSpace::triple<unsigned char> triple_t ;
triple_t rgb = Gr::ColourSpace::rgb_int( triple_t(y,u,v) ) ;
triple_t yuv = Gr::ColourSpace::yuv_int( triple_t(r,g,b) ) ;
triple_t rgb = Gr::ColourSpace::rgb( triple_t(y,u,v) ) ;
triple_t yuv = Gr::ColourSpace::yuv( triple_t(r,g,b) ) ;

Most code will be content to use just the two lookup functions rgb_int() and yuv_int(), and in that case the bulk of the code here is only used at built-time to generate the lookup tables that underlie those functions.

Otherwise, there are two top-level functions rgb_imp() and yuv_imp() with lots of template parameters providing maximum flexibility, and two convenience functions rgb() and yuv() that use rgb_imp() and yuv_imp() with sensible template defaults.

The implementation recognises that the core matrix multiplication has to be performed with a higher-precision numeric datatype than is used at the interface, and that the values used at the interface are sometimes constrained to a range that is not a convenient power of two (ie. head-room and toe-room for YUV values). The internal datatype can be considered to be "analogue" and the external datatype "digital", alluding to their different levels of precision.

The internal datatype is most naturally "double", but for speed on some machines it can be better to used a fixed-point type. Two fixed-point classes are provided, "fp_256" and "fp_1000", together with adaptors that are used to convert between internal and external values. If any internal datatype other than "double", "fp_256" or "fp_1000" is used then it will need to have a corresponding adaptor class (as per "converter_double").