interpolate8x8.cpp

一个用于智能手机的多媒体库适合S60 WinCE的跨平台开发库

/*****************************************************************************
 *
 * XVID MPEG-4 VIDEO CODEC
 * - 8x8 block-based halfpel interpolation -
 *
 *  Copyright(C) 2001-2003 Peter Ross <[e-mail]>
 *
 *  This program is free software ; you can redistribute it and/or modify
 *  it under the terms of the GNU General Public License as published by
 *  the Free Software Foundation ; either version 2 of the License, or
 *  (at your option) any later version.
 *
 *  This program is distributed in the hope that it will be useful,
 *  but WITHOUT ANY WARRANTY ; without even the implied warranty of
 *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 *  GNU General Public License for more details.
 *
 *  You should have received a copy of the GNU General Public License
 *  along with this program ; if not, write to the Free Software
 *  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307 USA
 *
 * $Id: interpolate8x8.cpp,v 1.1.1.1 2005/07/13 14:36:15 jeanlf Exp $
 *
 ****************************************************************************/

#include "portab.h"
#include "global.h"
#include "interpolate8x8.h"

//----------------------------

void interpolate8x8_switch(byte *cur, const byte *refn, dword x, dword y, int dx, int dy, dword stride, bool rounding){

   const byte *src = refn + ((y + (dy>>1)) * stride + x + (dx>>1));
   byte *dst = cur + (y * stride + x);

   switch(((dx & 1) << 1) + (dy & 1)){
   case 0:
      transfer8x8_copy(dst, src, stride);
      break;
   case 1:
      interpolate8x8_halfpel_v(dst, src, stride, rounding);
      break;
   case 2:
      interpolate8x8_halfpel_h(dst, src, stride, rounding);
      break;
   default:
      interpolate8x8_halfpel_hv(dst, src, stride, rounding);
      break;
   }
}

//----------------------------

void interpolate8x8_quarterpel(byte *cur, byte *refn, byte *refh, byte *refv, byte *refhv, dword x, dword y, int dx, int dy, dword stride, bool rounding){

   const int xRef = x*4 + dx;
   const int yRef = y*4 + dy;

   byte *src, *dst;
   byte *halfpel_h, *halfpel_v, *halfpel_hv;
   int x_int, y_int, x_frac, y_frac;

   x_int = xRef/4;
   if(xRef < 0 && xRef % 4)
      x_int--;

   x_frac = xRef - (4*x_int);

   y_int  = yRef/4;
   if (yRef < 0 && yRef % 4)
      y_int--;

   y_frac = yRef - (4*y_int);

   src = refn + y_int * stride + x_int;
   halfpel_h = refh;
   halfpel_v = refv;
   halfpel_hv = refhv;

   dst = cur + y * stride + x;

   switch((y_frac << 2) | (x_frac)) {

   case 0:
      transfer8x8_copy(dst, src, stride);
      break;

   case 1:
      interpolate8x8_lowpass_h(halfpel_h, src, stride, rounding);
      interpolate8x8_avg2(dst, src, halfpel_h, stride, rounding, 8);
      break;

   case 2:
       interpolate8x8_lowpass_h(dst, src, stride, rounding);
      break;

   case 3:
      interpolate8x8_lowpass_h(halfpel_h, src, stride, rounding);
      interpolate8x8_avg2(dst, src + 1, halfpel_h, stride, rounding, 8);
      break;

   case 4:
      interpolate8x8_lowpass_v(halfpel_v, src, stride, rounding);
      interpolate8x8_avg2(dst, src, halfpel_v, stride, rounding, 8);
      break;

   case 5:
      interpolate8x8_lowpass_h(halfpel_h, src, stride, rounding);
      interpolate8x8_avg2(halfpel_v, src, halfpel_h, stride, rounding, 9);
      interpolate8x8_lowpass_v(halfpel_hv, halfpel_v, stride, rounding);
      interpolate8x8_avg2(dst, halfpel_v, halfpel_hv, stride, rounding, 8);
      break;

   case 6:
      interpolate8x8_lowpass_hv(halfpel_hv, halfpel_h, src, stride, rounding);
      interpolate8x8_avg2(dst, halfpel_h, halfpel_hv, stride, rounding, 8);
      break;

   case 7:
      interpolate8x8_lowpass_h(halfpel_h, src, stride, rounding);
      interpolate8x8_avg2(halfpel_v, src + 1, halfpel_h, stride, rounding, 9);
      interpolate8x8_lowpass_v(halfpel_hv, halfpel_v, stride, rounding);
      interpolate8x8_avg2(dst, halfpel_v, halfpel_hv, stride, rounding, 8);
      break;

   case 8:
       interpolate8x8_lowpass_v(dst, src, stride, rounding);
      break;

   case 9:
      interpolate8x8_lowpass_h(halfpel_h, src, stride, rounding);
      interpolate8x8_avg2(halfpel_v, src, halfpel_h, stride, rounding, 9);
      interpolate8x8_lowpass_v(dst, halfpel_v, stride, rounding);
      break;

   case 10:
      interpolate8x8_lowpass_hv(dst, halfpel_h, src, stride, rounding);
      break;

   case 11:
      interpolate8x8_lowpass_h(halfpel_h, src, stride, rounding);
      interpolate8x8_avg2(halfpel_v, src + 1, halfpel_h, stride, rounding, 9);
      interpolate8x8_lowpass_v(dst, halfpel_v, stride, rounding);
         break;

   case 12:
      interpolate8x8_lowpass_v(halfpel_v, src, stride, rounding);
      interpolate8x8_avg2(dst, src+stride, halfpel_v, stride, rounding, 8);
         break;

   case 13:
      interpolate8x8_lowpass_h(halfpel_h, src, stride, rounding);
      interpolate8x8_avg2(halfpel_v, src, halfpel_h, stride, rounding, 9);
      interpolate8x8_lowpass_v(halfpel_hv, halfpel_v, stride, rounding);
      interpolate8x8_avg2(dst, halfpel_v+stride, halfpel_hv, stride, rounding, 8);
      break;

   case 14:
      interpolate8x8_lowpass_hv(halfpel_hv, halfpel_h, src, stride, rounding);
      interpolate8x8_avg2(dst, halfpel_h+stride, halfpel_hv, stride, rounding, 8);
      break;

   case 15:
      interpolate8x8_lowpass_h(halfpel_h, src, stride, rounding);
      interpolate8x8_avg2(halfpel_v, src + 1, halfpel_h, stride, rounding, 9);
      interpolate8x8_lowpass_v(halfpel_hv, halfpel_v, stride, rounding);
      interpolate8x8_avg2(dst, halfpel_hv, halfpel_v + stride, stride, rounding, 8);
      break;
   }
}

//----------------------------

void interpolate16x16_quarterpel(byte *cur, byte *refn, byte *refh, byte *refv, byte *refhv, dword x, dword y, int dx, int dy, dword stride, bool rounding){

   const int xRef = x*4 + dx;
   const int yRef = y*4 + dy;

   byte *src, *dst;
   byte *halfpel_h, *halfpel_v, *halfpel_hv;
   int x_int, y_int, x_frac, y_frac;

   x_int = xRef/4;
   if (xRef < 0 && xRef % 4)
      x_int--;

   x_frac = xRef - (4*x_int);

   y_int  = yRef/4;
   if (yRef < 0 && yRef % 4)
      y_int--;

   y_frac = yRef - (4*y_int);

   src = refn + y_int * stride + x_int;
   halfpel_h = refh;
   halfpel_v = refv;
   halfpel_hv = refhv;

   dst = cur + y * stride + x;

   switch((y_frac << 2) | (x_frac)){
   case 0:
      transfer16x16_copy(dst, src, stride);
      break;

   case 1:
      interpolate16x16_lowpass_h(halfpel_h, src, stride, rounding);
      interpolate8x8_avg2(dst, src, halfpel_h, stride, rounding, 8);
      interpolate8x8_avg2(dst+8, src+8, halfpel_h+8, stride, rounding, 8);
      interpolate8x8_avg2(dst+8*stride, src+8*stride, halfpel_h+8*stride, stride, rounding, 8);
      interpolate8x8_avg2(dst+8*stride+8, src+8*stride+8, halfpel_h+8*stride+8, stride, rounding, 8);
      break;

   case 2:
       interpolate16x16_lowpass_h(dst, src, stride, rounding);
      break;

   case 3:
      interpolate16x16_lowpass_h(halfpel_h, src, stride, rounding);
      interpolate8x8_avg2(dst, src + 1, halfpel_h, stride, rounding, 8);
      interpolate8x8_avg2(dst+8, src + 8 + 1, halfpel_h+8, stride, rounding, 8);
      interpolate8x8_avg2(dst+8*stride, src + 8*stride + 1, halfpel_h+8*stride, stride, rounding, 8);
      interpolate8x8_avg2(dst+8*stride+8, src+8*stride+8 + 1, halfpel_h+8*stride+8, stride, rounding, 8);
      break;

   case 4:
      interpolate16x16_lowpass_v(halfpel_v, src, stride, rounding);
      interpolate8x8_avg2(dst, src, halfpel_v, stride, rounding, 8);
      interpolate8x8_avg2(dst+8, src+8, halfpel_v+8, stride, rounding, 8);
      interpolate8x8_avg2(dst+8*stride, src+8*stride, halfpel_v+8*stride, stride, rounding, 8);
      interpolate8x8_avg2(dst+8*stride+8, src+8*stride+8, halfpel_v+8*stride+8, stride, rounding, 8);
      break;

   case 5:
      interpolate16x16_lowpass_h(halfpel_h, src, stride, rounding);
      interpolate8x8_avg2(halfpel_v, src, halfpel_h, stride, rounding, 9);
      interpolate8x8_avg2(halfpel_v+8, src + 8, halfpel_h+8, stride, rounding, 9);
      interpolate8x8_avg2(halfpel_v+8*stride, src + 8*stride, halfpel_h+8*stride, stride, rounding, 9);
      interpolate8x8_avg2(halfpel_v+8*stride+8, src+8*stride+8, halfpel_h+8*stride+8, stride, rounding, 9);

      interpolate16x16_lowpass_v(halfpel_hv, halfpel_v, stride, rounding);
      interpolate8x8_avg2(dst, halfpel_hv, halfpel_v, stride, rounding, 8);
      interpolate8x8_avg2(dst+8, halfpel_hv+8, halfpel_v+8, stride, rounding, 8);
      interpolate8x8_avg2(dst+8*stride, halfpel_hv+8*stride, halfpel_v+8*stride, stride, rounding, 8);
      interpolate8x8_avg2(dst+8*stride+8, halfpel_hv+8*stride+8, halfpel_v+8*stride+8, stride, rounding, 8);
      break;

   case 6:
      interpolate16x16_lowpass_hv(halfpel_hv, halfpel_h, src, stride, rounding);
      interpolate8x8_avg2(dst, halfpel_h, halfpel_hv, stride, rounding, 8);
      interpolate8x8_avg2(dst+8, halfpel_h+8, halfpel_hv+8, stride, rounding, 8);
      interpolate8x8_avg2(dst+8*stride, halfpel_h+8*stride, halfpel_hv+8*stride, stride, rounding, 8);
      interpolate8x8_avg2(dst+8*stride+8, halfpel_h+8*stride+8, halfpel_hv+8*stride+8, stride, rounding, 8);
      break;

   case 7:
      interpolate16x16_lowpass_h(halfpel_h, src, stride, rounding);
      interpolate8x8_avg2(halfpel_v, src+1, halfpel_h, stride, rounding, 9);
      interpolate8x8_avg2(halfpel_v+8, src+1 + 8, halfpel_h+8, stride, rounding, 9);
      interpolate8x8_avg2(halfpel_v+8*stride, src+1 + 8*stride, halfpel_h+8*stride, stride, rounding, 9);
      interpolate8x8_avg2(halfpel_v+8*stride+8, src+1+8*stride+8, halfpel_h+8*stride+8, stride, rounding, 9);

      interpolate16x16_lowpass_v(halfpel_hv, halfpel_v, stride, rounding);
      interpolate8x8_avg2(dst, halfpel_hv, halfpel_v, stride, rounding, 8);
      interpolate8x8_avg2(dst+8, halfpel_hv+8, halfpel_v+8, stride, rounding, 8);
      interpolate8x8_avg2(dst+8*stride, halfpel_hv+8*stride, halfpel_v+8*stride, stride, rounding, 8);
      interpolate8x8_avg2(dst+8*stride+8, halfpel_hv+8*stride+8, halfpel_v+8*stride+8, stride, rounding, 8);
      break;

   case 8:
       interpolate16x16_lowpass_v(dst, src, stride, rounding);
      break;

   case 9:
      interpolate16x16_lowpass_h(halfpel_h, src, stride, rounding);
      interpolate8x8_avg2(halfpel_v, src, halfpel_h, stride, rounding, 9);
      interpolate8x8_avg2(halfpel_v+8, src + 8, halfpel_h+8, stride, rounding, 9);
      interpolate8x8_avg2(halfpel_v+8*stride, src + 8*stride, halfpel_h+8*stride, stride, rounding, 9);
      interpolate8x8_avg2(halfpel_v+8*stride+8, src+8*stride+8, halfpel_h+8*stride+8, stride, rounding, 9);
      interpolate16x16_lowpass_v(dst, halfpel_v, stride, rounding);
      break;

   case 10:
      interpolate16x16_lowpass_hv(dst, halfpel_h, src, stride, rounding);
      break;

   case 11:
      interpolate16x16_lowpass_h(halfpel_h, src, stride, rounding);
      interpolate8x8_avg2(halfpel_v, src+1, halfpel_h, stride, rounding, 9);