sergiotarxz
/
libvips
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

actually, revert to the smooth limiter in LBB, which is best overall

This commit is contained in:
Nicolas Robidoux 2011-11-22 15:18:04 -05:00
parent b36717a070
commit b8d3c179f0
1 changed files with 126 additions and 136 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

258
libvips/resample/lbb.cpp
View File

@ -49,6 +49,31 @@
* allowance. * allowance.
*/ */
/*
* LBB has two versions:
*
* A "soft" version, which shows a little less staircasing and a
* little more haloing, and which is a little more expensive to
* compute. We recommend this as the default.
*
* A "sharp" version, which shows a little more staircasing and a
* little less haloing, which is a little cheaper (it uses 6 less
* comparisons and 12 less "? :").
*
* The only difference between the two is that the "soft" versions
* uses local minima and maxima computed over 3x3 square blocks, and
* the "sharp" version uses local minima and maxima computed over 3x3
* crosses.
*
* If you want to use the "sharp" version, comment out the following
* three pre-processor code lines:
*/
/*
#ifndef __LBB_CHEAP_H__
#define __LBB_CHEAP_H__
#endif
*/
/* /*
* LBB (Locally Bounded Bicubic) is a high quality nonlinear variant * LBB (Locally Bounded Bicubic) is a high quality nonlinear variant
* of Catmull-Rom. Images resampled with LBB have much smaller halos * of Catmull-Rom. Images resampled with LBB have much smaller halos
@ -70,9 +95,7 @@
* code was performed by C. Racette and N. Robidoux in the course of * code was performed by C. Racette and N. Robidoux in the course of
* her honours thesis, and by N. Robidoux, A. Turcotte and E. Daoust * her honours thesis, and by N. Robidoux, A. Turcotte and E. Daoust
* during Google Summer of Code 2009 (through two awards made to GIMP * during Google Summer of Code 2009 (through two awards made to GIMP
* to improve GEGL). The final version of LBB was formulated in * to improve GEGL).
* October 2011 by N. Robidoux based on insight gained while reviewing
* C. Racette's masters thesis.
* *
* LBB is a novel method with the following properties: * LBB is a novel method with the following properties:
* *
@ -127,9 +150,6 @@
* *
* The above paragraph described the "soft" version of LBB. The * The above paragraph described the "soft" version of LBB. The
* "sharp" version is similar. * "sharp" version is similar.
*
* A slightly different preliminary version of LBB is documented in
* C. Racette's masters thesis.
*/ */
#ifdef HAVE_CONFIG_H #ifdef HAVE_CONFIG_H
@ -178,7 +198,7 @@ typedef struct _VipsInterpolateLbbClass {
#define LBB_SIGN(x) ( ((x)>=0.) ? 1.0 : -1.0 ) #define LBB_SIGN(x) ( ((x)>=0.) ? 1.0 : -1.0 )
/* /*
* MIN and MAX macros set up so that I can put the likely winner in * MIN and MAX macros set up so that I can put the likely winner in
* the first argument (forward branch likely): * the first argument (forward branch likely blah blah blah):
*/ */
#define LBB_MIN(x,y) ( ((x)<=(y)) ? (x) : (y) ) #define LBB_MIN(x,y) ( ((x)<=(y)) ? (x) : (y) )
#define LBB_MAX(x,y) ( ((x)>=(y)) ? (x) : (y) ) #define LBB_MAX(x,y) ( ((x)>=(y)) ? (x) : (y) )
@ -238,95 +258,88 @@ lbbicubic( const double c00,
* where ix is the (pseudo-)floor of the requested left-to-right * where ix is the (pseudo-)floor of the requested left-to-right
* location ("X"), and iy is the floor of the requested up-to-down * location ("X"), and iy is the floor of the requested up-to-down
* location. * location.
*
* Below, "00", "10", "01" and "11" refer to the index "shifts" from
* the (ix,iy) position. That is,
*
* "00" refers to the dos_two position,
* "10" refers to the dos_thr position,
* "01" refers to the tre_two position, and
* "11" refers to the tre_thr position.
*/ */
#if defined (__LBB_CHEAP_H__)
/* /*
* Computation of the four pairs of horizontal min and max and four * Computation of the four min and four max over 3x3 input data
* pairs of vertical min and max over aligned groups of three input * sub-crosses of the 4x4 input stencil, performed with only 22
* pixel values, and four pairs of min and max over 3x3 input data * comparisons and 28 "? :". If you can figure out how to do this
* sub-blocks of the 4x4 input stencil. * more efficiently, let us know.
* *
* Cost: 48 conditional moves involving 42 comparisons. * This is the cheaper (but arguably less desirable in terms of
* quality) version of the computation.
*/ */
const double m1x = (dos_two <= dos_thr) ? dos_two : dos_thr ; const double m1 = (dos_two <= dos_thr) ? dos_two : dos_thr ;
const double M1x = (dos_two <= dos_thr) ? dos_thr : dos_two ; const double M1 = (dos_two <= dos_thr) ? dos_thr : dos_two ;
const double m2 = (tre_two <= tre_thr) ? tre_two : tre_thr ;
const double m2x = (tre_two <= tre_thr) ? tre_two : tre_thr ; const double M2 = (tre_two <= tre_thr) ? tre_thr : tre_two ;
const double M2x = (tre_two <= tre_thr) ? tre_thr : tre_two ; const double m3 = (uno_two <= dos_one) ? uno_two : dos_one ;
const double M3 = (uno_two <= dos_one) ? dos_one : uno_two ;
const double m1y = (dos_two <= tre_two) ? dos_two : tre_two ; const double m4 = (uno_thr <= dos_fou) ? uno_thr : dos_fou ;
const double M1y = (dos_two <= tre_two) ? tre_two : dos_two ; const double M4 = (uno_thr <= dos_fou) ? dos_fou : uno_thr ;
const double m5 = (tre_one <= qua_two) ? tre_one : qua_two ;
const double m2y = (dos_thr <= tre_thr) ? dos_thr : tre_thr ; const double M5 = (tre_one <= qua_two) ? qua_two : tre_one ;
const double M2y = (dos_thr <= tre_thr) ? tre_thr : dos_thr ; const double m6 = (tre_fou <= qua_thr) ? tre_fou : qua_thr ;
const double M6 = (tre_fou <= qua_thr) ? qua_thr : tre_fou ;
const double min00x = LBB_MIN( m1x, dos_one ); const double m7 = LBB_MIN( m1, tre_two );
const double max00x = LBB_MAX( M1x, dos_one ); const double M7 = LBB_MAX( M1, tre_two );
const double m8 = LBB_MIN( m1, tre_thr );
const double min10x = LBB_MIN( m1x, dos_fou ); const double M8 = LBB_MAX( M1, tre_thr );
const double max10x = LBB_MAX( M1x, dos_fou ); const double m9 = LBB_MIN( m2, dos_two );
const double M9 = LBB_MAX( M2, dos_two );
const double min01x = LBB_MIN( m2x, tre_one ); const double m10 = LBB_MIN( m2, dos_thr );
const double max01x = LBB_MAX( M2x, tre_one ); const double M10 = LBB_MAX( M2, dos_thr );
const double min00 = LBB_MIN( m7, m3 );
const double min11x = LBB_MIN( m2x, tre_fou ); const double max00 = LBB_MAX( M7, M3 );
const double max11x = LBB_MAX( M2x, tre_fou ); const double min10 = LBB_MIN( m8, m4 );
const double max10 = LBB_MAX( M8, M4 );
const double min00y = LBB_MIN( m1y, uno_two ); const double min01 = LBB_MIN( m9, m5 );
const double max00y = LBB_MAX( M1y, uno_two ); const double max01 = LBB_MAX( M9, M5 );
const double min11 = LBB_MIN( m10, m6 );
const double min10y = LBB_MIN( m1y, qua_two ); const double max11 = LBB_MAX( M10, M6 );
const double max10y = LBB_MAX( M1y, qua_two ); #else
/*
const double min01y = LBB_MIN( m2y, uno_thr ); * Computation of the four min and four max over 3x3 input data
const double max01y = LBB_MAX( M2y, uno_thr ); * sub-blocks of the 4x4 input stencil, performed with only 28
* comparisons and 34 "? :". If you can figure how to do this more
const double min11y = LBB_MIN( m2y, qua_thr ); * efficiently, let us know.
const double max11y = LBB_MAX( M2y, qua_thr ); */
const double m1 = (dos_two <= dos_thr) ? dos_two : dos_thr ;
const double m3x = (uno_two <= uno_thr) ? uno_two : uno_thr ; const double M1 = (dos_two <= dos_thr) ? dos_thr : dos_two ;
const double M3x = (uno_two <= uno_thr) ? uno_thr : uno_two ; const double m2 = (tre_two <= tre_thr) ? tre_two : tre_thr ;
const double M2 = (tre_two <= tre_thr) ? tre_thr : tre_two ;
const double m4x = (qua_two <= qua_thr) ? qua_two : qua_thr ; const double m6 = (dos_one <= tre_one) ? dos_one : tre_one ;
const double M4x = (qua_two <= qua_thr) ? qua_thr : qua_two ; const double M6 = (dos_one <= tre_one) ? tre_one : dos_one ;
const double m7 = (dos_fou <= tre_fou) ? dos_fou : tre_fou ;
const double m5x = LBB_MIN( m3x, uno_one ); const double M7 = (dos_fou <= tre_fou) ? tre_fou : dos_fou ;
const double M5x = LBB_MAX( M3x, uno_one ); const double m3 = (uno_two <= uno_thr) ? uno_two : uno_thr ;
const double M3 = (uno_two <= uno_thr) ? uno_thr : uno_two ;
const double m6x = LBB_MIN( m3x, uno_fou ); const double m4 = (qua_two <= qua_thr) ? qua_two : qua_thr ;
const double M6x = LBB_MAX( M3x, uno_fou ); const double M4 = (qua_two <= qua_thr) ? qua_thr : qua_two ;
const double m5 = LBB_MIN( m1, m2 );
const double m7x = LBB_MIN( m4x, qua_one ); const double M5 = LBB_MAX( M1, M2 );
const double M7x = LBB_MAX( M4x, qua_one ); const double m10 = LBB_MIN( m6, uno_one );
const double M10 = LBB_MAX( M6, uno_one );
const double m8x = LBB_MIN( m4x, qua_fou ); const double m11 = LBB_MIN( m6, qua_one );
const double M8x = LBB_MAX( M4x, qua_fou ); const double M11 = LBB_MAX( M6, qua_one );
const double m12 = LBB_MIN( m7, uno_fou );
const double m3y = LBB_MIN( min00x, min01x ); const double M12 = LBB_MAX( M7, uno_fou );
const double M3y = LBB_MAX( max00x, max01x ); const double m13 = LBB_MIN( m7, qua_fou );
const double M13 = LBB_MAX( M7, qua_fou );
const double m4y = LBB_MIN( min10x, min11x ); const double m8 = LBB_MIN( m5, m3 );
const double M4y = LBB_MAX( max10x, max10x ); const double M8 = LBB_MAX( M5, M3 );
const double m9 = LBB_MIN( m5, m4 );
const double min00 = LBB_MIN( m3y, m5x ); const double M9 = LBB_MAX( M5, M4 );
const double max00 = LBB_MAX( M3y, M5x ); const double min00 = LBB_MIN( m8, m10 );
const double max00 = LBB_MAX( M8, M10 );
const double min10 = LBB_MIN( m4y, m6x ); const double min10 = LBB_MIN( m8, m12 );
const double max10 = LBB_MAX( M4y, M6x ); const double max10 = LBB_MAX( M8, M12 );
const double min01 = LBB_MIN( m9, m11 );
const double min01 = LBB_MIN( m3y, m7x ); const double max01 = LBB_MAX( M9, M11 );
const double max01 = LBB_MAX( M3y, M7x ); const double min11 = LBB_MIN( m9, m13 );
const double max11 = LBB_MAX( M9, M13 );
const double min11 = LBB_MIN( m4y, m8x ); #endif
const double max11 = LBB_MAX( M4y, M8x );
/* /*
* The remainder of the "per channel" computation involves the * The remainder of the "per channel" computation involves the
@ -354,24 +367,6 @@ lbbicubic( const double c00,
/* /*
* Distances to the local min and max: * Distances to the local min and max:
*/ */
const double u00x = dos_two - min00x;
const double v00x = max00x - dos_two;
const double u10x = dos_thr - min10x;
const double v10x = max10x - dos_thr;
const double u01x = tre_two - min01x;
const double v01x = max01x - tre_two;
const double u11x = tre_thr - min11x;
const double v11x = max11x - tre_thr;
const double u00y = dos_two - min00y;
const double v00y = max00y - dos_two;
const double u10y = dos_thr - min10y;
const double v10y = max10y - dos_thr;
const double u01y = tre_two - min01y;
const double v01y = max01y - tre_two;
const double u11y = tre_thr - min11y;
const double v11y = max11y - tre_thr;
const double u00 = dos_two - min00; const double u00 = dos_two - min00;
const double v00 = max00 - dos_two; const double v00 = max00 - dos_two;
const double u10 = dos_thr - min10; const double u10 = dos_thr - min10;
@ -423,43 +418,38 @@ lbbicubic( const double c00,
* Slope limiters. The key multiplier is 3 but we fold a factor of * Slope limiters. The key multiplier is 3 but we fold a factor of
* 2, hence 6: * 2, hence 6:
*/ */
const double dble_slopelimit_00x = 6.0 * LBB_MIN( u00x, v00x ); const double dble_slopelimit_00 = 6.0 * LBB_MIN( u00, v00 );
const double dble_slopelimit_10x = 6.0 * LBB_MIN( u10x, v10x ); const double dble_slopelimit_10 = 6.0 * LBB_MIN( u10, v10 );
const double dble_slopelimit_01x = 6.0 * LBB_MIN( u01x, v01x ); const double dble_slopelimit_01 = 6.0 * LBB_MIN( u01, v01 );
const double dble_slopelimit_11x = 6.0 * LBB_MIN( u11x, v11x ); const double dble_slopelimit_11 = 6.0 * LBB_MIN( u11, v11 );
const double dble_slopelimit_00y = 6.0 * LBB_MIN( u00y, v00y );
const double dble_slopelimit_10y = 6.0 * LBB_MIN( u10y, v10y );
const double dble_slopelimit_01y = 6.0 * LBB_MIN( u01y, v01y );
const double dble_slopelimit_11y = 6.0 * LBB_MIN( u11y, v11y );
/* /*
* Clamped first derivatives: * Clamped first derivatives:
*/ */
const double dble_dzdx00 = const double dble_dzdx00 =
( sign_dzdx00 * dble_dzdx00i <= dble_slopelimit_00x ) ( sign_dzdx00 * dble_dzdx00i <= dble_slopelimit_00 )
? dble_dzdx00i : sign_dzdx00 * dble_slopelimit_00x; ? dble_dzdx00i : sign_dzdx00 * dble_slopelimit_00;
const double dble_dzdy00 = const double dble_dzdy00 =
( sign_dzdy00 * dble_dzdy00i <= dble_slopelimit_00y ) ( sign_dzdy00 * dble_dzdy00i <= dble_slopelimit_00 )
? dble_dzdy00i : sign_dzdy00 * dble_slopelimit_00y; ? dble_dzdy00i : sign_dzdy00 * dble_slopelimit_00;
const double dble_dzdx10 = const double dble_dzdx10 =
( sign_dzdx10 * dble_dzdx10i <= dble_slopelimit_10x ) ( sign_dzdx10 * dble_dzdx10i <= dble_slopelimit_10 )
? dble_dzdx10i : sign_dzdx10 * dble_slopelimit_10x; ? dble_dzdx10i : sign_dzdx10 * dble_slopelimit_10;
const double dble_dzdy10 = const double dble_dzdy10 =
( sign_dzdy10 * dble_dzdy10i <= dble_slopelimit_10y ) ( sign_dzdy10 * dble_dzdy10i <= dble_slopelimit_10 )
? dble_dzdy10i : sign_dzdy10 * dble_slopelimit_10y; ? dble_dzdy10i : sign_dzdy10 * dble_slopelimit_10;
const double dble_dzdx01 = const double dble_dzdx01 =
( sign_dzdx01 * dble_dzdx01i <= dble_slopelimit_01x ) ( sign_dzdx01 * dble_dzdx01i <= dble_slopelimit_01 )
? dble_dzdx01i : sign_dzdx01 * dble_slopelimit_01x; ? dble_dzdx01i : sign_dzdx01 * dble_slopelimit_01;
const double dble_dzdy01 = const double dble_dzdy01 =
( sign_dzdy01 * dble_dzdy01i <= dble_slopelimit_01y ) ( sign_dzdy01 * dble_dzdy01i <= dble_slopelimit_01 )
? dble_dzdy01i : sign_dzdy01 * dble_slopelimit_01y; ? dble_dzdy01i : sign_dzdy01 * dble_slopelimit_01;
const double dble_dzdx11 = const double dble_dzdx11 =
( sign_dzdx11 * dble_dzdx11i <= dble_slopelimit_11x ) ( sign_dzdx11 * dble_dzdx11i <= dble_slopelimit_11 )
? dble_dzdx11i : sign_dzdx11 * dble_slopelimit_11x; ? dble_dzdx11i : sign_dzdx11 * dble_slopelimit_11;
const double dble_dzdy11 = const double dble_dzdy11 =
( sign_dzdy11 * dble_dzdy11i <= dble_slopelimit_11y ) ( sign_dzdy11 * dble_dzdy11i <= dble_slopelimit_11 )
? dble_dzdy11i : sign_dzdy11 * dble_slopelimit_11y; ? dble_dzdy11i : sign_dzdy11 * dble_slopelimit_11;
/* /*
* Sums and differences of first derivatives: * Sums and differences of first derivatives: