generic-library/vpx
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity
aea29cd278
vpx/vp9/common/vp9_reconintra4x4.c
Yaowu Xu 3d655805f2 Use same intra prediction for all block size 
The commmit changed to use same intra prediction function for all
block sizes.

Some details on the changes:
1. All directional modes except DC/TM/V/H now have built-in filtering
for all pixels with filter taps either (1, 2, 1)/4 or (1, 1)/2.
2. Above edge get automatic extended to double width (bw*2), which
makes a lot of the prediciton mode computation simpler.
3. Same intra prediction function is called with different size
for i4x4_pred and all other larger size.

Overall, the change helped keyframe only coding for both cif size
and std-hd size test sets by .5% consistently on all encodings.
For normal coding with single/auto key frame, the change now also
is consistently net positive for all encodings. The overall gains
is about .15% on std-hd set.

Change-Id: I01ceb31fbc73d49776262e6bdc06853b03bbd1d1
2013-04-29 15:15:30 -07:00

452 lines
15 KiB
C
Raw Blame History

/*
* Copyright (c) 2010 The WebM project authors. All Rights Reserved.
*
* Use of this source code is governed by a BSD-style license
* that can be found in the LICENSE file in the root of the source
* tree. An additional intellectual property rights grant can be found
* in the file PATENTS. All contributing project authors may
* be found in the AUTHORS file in the root of the source tree.
*/
#include "./vpx_config.h"
#include "vpx_mem/vpx_mem.h"
#include "vp9/common/vp9_reconintra.h"
#include "vp9_rtcd.h"
#if CONFIG_NEWBINTRAMODES
static int find_grad_measure(uint8_t *x, int stride, int n, int tx, int ty,
int dx, int dy) {
int i, j;
int count = 0, gsum = 0, gdiv;
/* TODO: Make this code more efficient by breaking up into two loops */
for (i = -ty; i < n; ++i)
for (j = -tx; j < n; ++j) {
int g;
if (i >= 0 && j >= 0) continue;
if (i + dy >= 0 && j + dx >= 0) continue;
if (i + dy < -ty || i + dy >= n || j + dx < -tx || j + dx >= n) continue;
g = abs(x[(i + dy) * stride + j + dx] - x[i * stride + j]);
gsum += g * g;
count++;
}
gdiv = (dx * dx + dy * dy) * count;
return ((gsum << 8) + (gdiv >> 1)) / gdiv;
}
#if CONTEXT_PRED_REPLACEMENTS == 6
B_PREDICTION_MODE vp9_find_dominant_direction(uint8_t *ptr,
int stride, int n,
int tx, int ty) {
int g[8], i, imin, imax;
g[1] = find_grad_measure(ptr, stride, n, tx, ty, 2, 1);
g[2] = find_grad_measure(ptr, stride, n, tx, ty, 1, 1);
g[3] = find_grad_measure(ptr, stride, n, tx, ty, 1, 2);
g[5] = find_grad_measure(ptr, stride, n, tx, ty, -1, 2);
g[6] = find_grad_measure(ptr, stride, n, tx, ty, -1, 1);
g[7] = find_grad_measure(ptr, stride, n, tx, ty, -2, 1);
imin = 1;
for (i = 2; i < 8; i += 1 + (i == 3))
imin = (g[i] < g[imin] ? i : imin);
imax = 1;
for (i = 2; i < 8; i += 1 + (i == 3))
imax = (g[i] > g[imax] ? i : imax);
/*
printf("%d %d %d %d %d %d = %d %d\n",
g[1], g[2], g[3], g[5], g[6], g[7], imin, imax);
*/
switch (imin) {
case 1:
return B_D153_PRED;
case 2:
return B_D135_PRED;
case 3:
return B_D117_PRED;
case 5:
return B_D63_PRED;
case 6:
return B_D45_PRED;
case 7:
return B_D27_PRED;
default:
assert(0);
}
}
#elif CONTEXT_PRED_REPLACEMENTS == 4
B_PREDICTION_MODE vp9_find_dominant_direction(uint8_t *ptr,
int stride, int n,
int tx, int ty) {
int g[8], i, imin, imax;
g[1] = find_grad_measure(ptr, stride, n, tx, ty, 2, 1);
g[3] = find_grad_measure(ptr, stride, n, tx, ty, 1, 2);
g[5] = find_grad_measure(ptr, stride, n, tx, ty, -1, 2);
g[7] = find_grad_measure(ptr, stride, n, tx, ty, -2, 1);
imin = 1;
for (i = 3; i < 8; i+=2)
imin = (g[i] < g[imin] ? i : imin);
imax = 1;
for (i = 3; i < 8; i+=2)
imax = (g[i] > g[imax] ? i : imax);
/*
printf("%d %d %d %d = %d %d\n",
g[1], g[3], g[5], g[7], imin, imax);
*/
switch (imin) {
case 1:
return B_D153_PRED;
case 3:
return B_D117_PRED;
case 5:
return B_D63_PRED;
case 7:
return B_D27_PRED;
default:
assert(0);
}
}
#elif CONTEXT_PRED_REPLACEMENTS == 0
B_PREDICTION_MODE vp9_find_dominant_direction(uint8_t *ptr,
int stride, int n,
int tx, int ty) {
int g[8], i, imin, imax;
g[0] = find_grad_measure(ptr, stride, n, tx, ty, 1, 0);
g[1] = find_grad_measure(ptr, stride, n, tx, ty, 2, 1);
g[2] = find_grad_measure(ptr, stride, n, tx, ty, 1, 1);
g[3] = find_grad_measure(ptr, stride, n, tx, ty, 1, 2);
g[4] = find_grad_measure(ptr, stride, n, tx, ty, 0, 1);
g[5] = find_grad_measure(ptr, stride, n, tx, ty, -1, 2);
g[6] = find_grad_measure(ptr, stride, n, tx, ty, -1, 1);
g[7] = find_grad_measure(ptr, stride, n, tx, ty, -2, 1);
imax = 0;
for (i = 1; i < 8; i++)
imax = (g[i] > g[imax] ? i : imax);
imin = 0;
for (i = 1; i < 8; i++)
imin = (g[i] < g[imin] ? i : imin);
switch (imin) {
case 0:
return B_H_PRED;
case 1:
return B_D153_PRED;
case 2:
return B_D135_PRED;
case 3:
return B_D117_PRED;
case 4:
return B_V_PRED;
case 5:
return B_D63_PRED;
case 6:
return B_D45_PRED;
case 7:
return B_D27_PRED;
default:
assert(0);
}
}
#endif
B_PREDICTION_MODE vp9_find_bpred_context(MACROBLOCKD *xd, int block_idx,
uint8_t *ptr, int stride) {
const int have_top = (block_idx >> 2) || xd->up_available;
const int have_left = (block_idx & 3) || xd->left_available;
int tx = have_left ? 4 : 0;
int ty = have_top ? 4 : 0;
if (!have_left && !have_top)
return B_DC_PRED;
return vp9_find_dominant_direction(ptr, stride, 4, tx, ty);
}
void vp9_intra4x4_predict(MACROBLOCKD *xd,
int block_idx,
int b_mode,
uint8_t *predictor,
int ps) {
int i, r, c;
const int have_top = (block_idx >> 2) || xd->up_available;
const int have_left = (block_idx & 3) || xd->left_available;
const int have_right = (block_idx & 3) != 3 || xd->right_available;
uint8_t left[4], above[8], top_left;
/*
* 127 127 127 .. 127 127 127 127 127 127
* 129 A B .. Y Z
* 129 C D .. W X
* 129 E F .. U V
* 129 G H .. S T T T T T
* ..
*/
if (have_left) {
uint8_t *left_ptr = predictor - 1;
const int stride = ps;
left[0] = left_ptr[0 * stride];
left[1] = left_ptr[1 * stride];
left[2] = left_ptr[2 * stride];
left[3] = left_ptr[3 * stride];
} else {
left[0] = left[1] = left[2] = left[3] = 129;
}
if (have_top) {
uint8_t *above_ptr = predictor - ps;
top_left = have_left ? above_ptr[-1] : 127;
above[0] = above_ptr[0];
above[1] = above_ptr[1];
above[2] = above_ptr[2];
above[3] = above_ptr[3];
if (((block_idx & 3) != 3) ||
(have_right && block_idx == 3 &&
((xd->mb_index != 3 && xd->sb_index != 3) ||
((xd->mb_index & 1) == 0 && xd->sb_index == 3)))) {
above[4] = above_ptr[4];
above[5] = above_ptr[5];
above[6] = above_ptr[6];
above[7] = above_ptr[7];
} else if (have_right) {
uint8_t *above_right = above_ptr + 4;
if (xd->sb_index == 3 && (xd->mb_index & 1))
above_right -= 32 * ps;
if (xd->mb_index == 3)
above_right -= 16 * ps;
above_right -= (block_idx & ~3) * ps;
/* use a more distant above-right (from closest available top-right
* corner), but with a "localized DC" (similar'ish to TM-pred):
*
* A B C D E F G H
* I J K L
* M N O P
* Q R S T
* U V W X x1 x2 x3 x4
*
* Where:
* x1 = clip_pixel(E + X - D)
* x2 = clip_pixel(F + X - D)
* x3 = clip_pixel(G + X - D)
* x4 = clip_pixel(H + X - D)
*
* This is applied anytime when we use a "distant" above-right edge
* that is not immediately top-right to the block that we're going
* to do intra prediction for.
*/
above[4] = clip_pixel(above_right[0] + above_ptr[3] - above_right[-1]);
above[5] = clip_pixel(above_right[1] + above_ptr[3] - above_right[-1]);
above[6] = clip_pixel(above_right[2] + above_ptr[3] - above_right[-1]);
above[7] = clip_pixel(above_right[3] + above_ptr[3] - above_right[-1]);
} else {
// extend edge
above[4] = above[5] = above[6] = above[7] = above[3];
}
} else {
above[0] = above[1] = above[2] = above[3] = 127;
above[4] = above[5] = above[6] = above[7] = 127;
top_left = 127;
}
#if CONFIG_NEWBINTRAMODES
if (b_mode == B_CONTEXT_PRED)
b_mode = xd->mode_info_context->bmi[block_idx].as_mode.context;
#endif
switch (b_mode) {
case B_DC_PRED: {
int expected_dc = 128;
if (have_top || have_left) {
int average = 0;
int count = 0;
if (have_top) {
for (i = 0; i < 4; i++)
average += above[i];
count += 4;
}
if (have_left) {
for (i = 0; i < 4; i++)
average += left[i];
count += 4;
}
expected_dc = (average + (count >> 1)) / count;
}
for (r = 0; r < 4; r++) {
for (c = 0; c < 4; c++)
predictor[c] = expected_dc;
predictor += ps;
}
}
break;
case B_TM_PRED: {
/* prediction similar to true_motion prediction */
for (r = 0; r < 4; r++) {
for (c = 0; c < 4; c++)
predictor[c] = clip_pixel(above[c] - top_left + left[r]);
predictor += ps;
}
}
break;
case B_V_PRED:
for (r = 0; r < 4; r++) {
for (c = 0; c < 4; c++)
predictor[c] = above[c];
predictor += ps;
}
break;
case B_H_PRED:
for (r = 0; r < 4; r++) {
for (c = 0; c < 4; c++)
predictor[c] = left[r];
predictor += ps;
}
break;
case B_D45_PRED: {
uint8_t *p = above;
predictor[0 * ps + 0] = ROUND_POWER_OF_TWO(p[0] + p[1] * 2 + p[2], 2);
predictor[0 * ps + 1] =
predictor[1 * ps + 0] = ROUND_POWER_OF_TWO(p[1] + p[2] * 2 + p[3], 2);
predictor[0 * ps + 2] =
predictor[1 * ps + 1] =
predictor[2 * ps + 0] = ROUND_POWER_OF_TWO(p[2] + p[3] * 2 + p[4], 2);
predictor[0 * ps + 3] =
predictor[1 * ps + 2] =
predictor[2 * ps + 1] =
predictor[3 * ps + 0] =
ROUND_POWER_OF_TWO(p[3] + p[4] * 2 + p[5], 2);
predictor[1 * ps + 3] =
predictor[2 * ps + 2] =
predictor[3 * ps + 1] = ROUND_POWER_OF_TWO(p[4] + p[5] * 2 + p[6], 2);
predictor[2 * ps + 3] =
predictor[3 * ps + 2] = ROUND_POWER_OF_TWO(p[5] + p[6] * 2 + p[7], 2);
predictor[3 * ps + 3] = ROUND_POWER_OF_TWO(p[6] + p[7] * 2 + p[7], 2);
}
break;
case B_D135_PRED: {
uint8_t p[9] = { left[3], left[2], left[1], left[0],
top_left,
above[0], above[1], above[2], above[3] };
predictor[3 * ps + 0] = ROUND_POWER_OF_TWO(p[0] + p[1] * 2 + p[2], 2);
predictor[3 * ps + 1] =
predictor[2 * ps + 0] = ROUND_POWER_OF_TWO(p[1] + p[2] * 2 + p[3], 2);
predictor[3 * ps + 2] =
predictor[2 * ps + 1] =
predictor[1 * ps + 0] = ROUND_POWER_OF_TWO(p[2] + p[3] * 2 + p[4], 2);
predictor[3 * ps + 3] =
predictor[2 * ps + 2] =
predictor[1 * ps + 1] =
predictor[0 * ps + 0] =
ROUND_POWER_OF_TWO(p[3] + p[4] * 2 + p[5], 2);
predictor[2 * ps + 3] =
predictor[1 * ps + 2] =
predictor[0 * ps + 1] = ROUND_POWER_OF_TWO(p[4] + p[5] * 2 + p[6], 2);
predictor[1 * ps + 3] =
predictor[0 * ps + 2] = ROUND_POWER_OF_TWO(p[5] + p[6] * 2 + p[7], 2);
predictor[0 * ps + 3] = ROUND_POWER_OF_TWO(p[6] + p[7] * 2 + p[8], 2);
}
break;
case B_D117_PRED: {
uint8_t p[9] = { left[3], left[2], left[1], left[0],
top_left,
above[0], above[1], above[2], above[3] };
predictor[3 * ps + 0] = ROUND_POWER_OF_TWO(p[1] + p[2] * 2 + p[3], 2);
predictor[2 * ps + 0] = ROUND_POWER_OF_TWO(p[2] + p[3] * 2 + p[4], 2);
predictor[3 * ps + 1] =
predictor[1 * ps + 0] = ROUND_POWER_OF_TWO(p[3] + p[4] * 2 + p[5], 2);
predictor[2 * ps + 1] =
predictor[0 * ps + 0] = ROUND_POWER_OF_TWO(p[4] + p[5], 1);
predictor[3 * ps + 2] =
predictor[1 * ps + 1] = ROUND_POWER_OF_TWO(p[4] + p[5] * 2 + p[6], 2);
predictor[2 * ps + 2] =
predictor[0 * ps + 1] = ROUND_POWER_OF_TWO(p[5] + p[6], 1);
predictor[3 * ps + 3] =
predictor[1 * ps + 2] = ROUND_POWER_OF_TWO(p[5] + p[6] * 2 + p[7], 2);
predictor[2 * ps + 3] =
predictor[0 * ps + 2] = ROUND_POWER_OF_TWO(p[6] + p[7], 1);
predictor[1 * ps + 3] = ROUND_POWER_OF_TWO(p[6] + p[7] * 2 + p[8], 2);
predictor[0 * ps + 3] = ROUND_POWER_OF_TWO(p[7] + p[8], 1);
}
break;
case B_D63_PRED: {
uint8_t *p = above;
predictor[0 * ps + 0] = ROUND_POWER_OF_TWO(p[0] + p[1], 1);
predictor[1 * ps + 0] = ROUND_POWER_OF_TWO(p[0] + p[1] * 2 + p[2], 2);
predictor[2 * ps + 0] =
predictor[0 * ps + 1] = ROUND_POWER_OF_TWO(p[1] + p[2], 1);
predictor[1 * ps + 1] =
predictor[3 * ps + 0] = ROUND_POWER_OF_TWO(p[1] + p[2] * 2 + p[3], 2);
predictor[2 * ps + 1] =
predictor[0 * ps + 2] = ROUND_POWER_OF_TWO(p[2] + p[3], 1);
predictor[3 * ps + 1] =
predictor[1 * ps + 2] = ROUND_POWER_OF_TWO(p[2] + p[3] * 2 + p[4], 2);
predictor[0 * ps + 3] =
predictor[2 * ps + 2] = ROUND_POWER_OF_TWO(p[3] + p[4], 1);
predictor[1 * ps + 3] =
predictor[3 * ps + 2] = ROUND_POWER_OF_TWO(p[3] + p[4] * 2 + p[5], 2);
predictor[2 * ps + 3] = ROUND_POWER_OF_TWO(p[4] + p[5] * 2 + p[6], 2);
predictor[3 * ps + 3] = ROUND_POWER_OF_TWO(p[5] + p[6] * 2 + p[7], 2);
}
break;
case B_D153_PRED: {
uint8_t p[9] = { left[3], left[2], left[1], left[0],
top_left,
above[0], above[1], above[2], above[3] };
predictor[3 * ps + 0] = ROUND_POWER_OF_TWO(p[0] + p[1], 1);
predictor[3 * ps + 1] = ROUND_POWER_OF_TWO(p[0] + p[1] * 2 + p[2], 2);
predictor[2 * ps + 0] =
predictor[3 * ps + 2] = ROUND_POWER_OF_TWO(p[1] + p[2], 1);
predictor[2 * ps + 1] =
predictor[3 * ps + 3] = ROUND_POWER_OF_TWO(p[1] + p[2] * 2 + p[3], 2);
predictor[2 * ps + 2] =
predictor[1 * ps + 0] = ROUND_POWER_OF_TWO(p[2] + p[3], 1);
predictor[2 * ps + 3] =
predictor[1 * ps + 1] = ROUND_POWER_OF_TWO(p[2] + p[3] * 2 + p[4], 2);
predictor[1 * ps + 2] =
predictor[0 * ps + 0] = ROUND_POWER_OF_TWO(p[3] + p[4], 1);
predictor[1 * ps + 3] =
predictor[0 * ps + 1] = ROUND_POWER_OF_TWO(p[3] + p[4] * 2 + p[5], 2);
predictor[0 * ps + 2] = ROUND_POWER_OF_TWO(p[4] + p[5] * 2 + p[6], 2);
predictor[0 * ps + 3] = ROUND_POWER_OF_TWO(p[5] + p[6] * 2 + p[7], 2);
}
break;
case B_D27_PRED: {
uint8_t *p = left;
predictor[0 * ps + 0] = ROUND_POWER_OF_TWO(p[0] + p[1], 1);
predictor[0 * ps + 1] = ROUND_POWER_OF_TWO(p[0] + p[1] * 2 + p[2], 2);
predictor[0 * ps + 2] =
predictor[1 * ps + 0] = ROUND_POWER_OF_TWO(p[1] + p[2], 1);
predictor[0 * ps + 3] =
predictor[1 * ps + 1] = ROUND_POWER_OF_TWO(p[1] + p[2] * 2 + p[3], 2);
predictor[1 * ps + 2] =
predictor[2 * ps + 0] = ROUND_POWER_OF_TWO(p[2] + p[3], 1);
predictor[1 * ps + 3] =
predictor[2 * ps + 1] = ROUND_POWER_OF_TWO(p[2] + p[3] * 2 + p[3], 2);
predictor[2 * ps + 2] =
predictor[2 * ps + 3] =
predictor[3 * ps + 0] =
predictor[3 * ps + 1] =
predictor[3 * ps + 2] =
predictor[3 * ps + 3] = p[3];
}
break;
#if CONFIG_NEWBINTRAMODES
case B_CONTEXT_PRED:
break;
/*
case B_CORNER_PRED:
corner_predictor(predictor, 16, 4, above, left);
break;
*/
#endif
}
}
#endif
Reference in New Issue View Git Blame Copy Permalink