2016-09-02 22:04:54 +03:00
|
|
|
/*
|
|
|
|
* Copyright (c) 2016, Alliance for Open Media. All rights reserved
|
|
|
|
*
|
|
|
|
* This source code is subject to the terms of the BSD 2 Clause License and
|
|
|
|
* the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License
|
|
|
|
* was not distributed with this source code in the LICENSE file, you can
|
|
|
|
* obtain it at www.aomedia.org/license/software. If the Alliance for Open
|
|
|
|
* Media Patent License 1.0 was not distributed with this source code in the
|
|
|
|
* PATENTS file, you can obtain it at www.aomedia.org/license/patent.
|
|
|
|
*/
|
2016-08-15 20:27:19 +03:00
|
|
|
#ifdef HAVE_CONFIG_H
|
2016-08-19 01:10:22 +03:00
|
|
|
#include "config.h"
|
2016-08-15 20:27:19 +03:00
|
|
|
#endif
|
|
|
|
|
2016-10-08 01:10:19 +03:00
|
|
|
// clang-format off
|
|
|
|
|
2016-08-15 20:27:19 +03:00
|
|
|
#include <stdlib.h>
|
|
|
|
#include <math.h>
|
|
|
|
#include "dering.h"
|
2016-10-07 00:27:34 +03:00
|
|
|
#include "./av1_rtcd.h"
|
2016-08-15 20:27:19 +03:00
|
|
|
|
|
|
|
/* Generated from gen_filter_tables.c. */
|
|
|
|
const int OD_DIRECTION_OFFSETS_TABLE[8][3] = {
|
2016-08-19 01:10:22 +03:00
|
|
|
{ -1 * OD_FILT_BSTRIDE + 1, -2 * OD_FILT_BSTRIDE + 2,
|
|
|
|
-3 * OD_FILT_BSTRIDE + 3 },
|
|
|
|
{ 0 * OD_FILT_BSTRIDE + 1, -1 * OD_FILT_BSTRIDE + 2,
|
|
|
|
-1 * OD_FILT_BSTRIDE + 3 },
|
|
|
|
{ 0 * OD_FILT_BSTRIDE + 1, 0 * OD_FILT_BSTRIDE + 2, 0 * OD_FILT_BSTRIDE + 3 },
|
|
|
|
{ 0 * OD_FILT_BSTRIDE + 1, 1 * OD_FILT_BSTRIDE + 2, 1 * OD_FILT_BSTRIDE + 3 },
|
|
|
|
{ 1 * OD_FILT_BSTRIDE + 1, 2 * OD_FILT_BSTRIDE + 2, 3 * OD_FILT_BSTRIDE + 3 },
|
|
|
|
{ 1 * OD_FILT_BSTRIDE + 0, 2 * OD_FILT_BSTRIDE + 1, 3 * OD_FILT_BSTRIDE + 1 },
|
|
|
|
{ 1 * OD_FILT_BSTRIDE + 0, 2 * OD_FILT_BSTRIDE + 0, 3 * OD_FILT_BSTRIDE + 0 },
|
|
|
|
{ 1 * OD_FILT_BSTRIDE + 0, 2 * OD_FILT_BSTRIDE - 1, 3 * OD_FILT_BSTRIDE - 1 },
|
2016-08-15 20:27:19 +03:00
|
|
|
};
|
|
|
|
|
|
|
|
/* Detect direction. 0 means 45-degree up-right, 2 is horizontal, and so on.
|
|
|
|
The search minimizes the weighted variance along all the lines in a
|
|
|
|
particular direction, i.e. the squared error between the input and a
|
|
|
|
"predicted" block where each pixel is replaced by the average along a line
|
|
|
|
in a particular direction. Since each direction have the same sum(x^2) term,
|
|
|
|
that term is never computed. See Section 2, step 2, of:
|
|
|
|
http://jmvalin.ca/notes/intra_paint.pdf */
|
2016-10-07 00:27:34 +03:00
|
|
|
int od_dir_find8_c(const od_dering_in *img, int stride, int32_t *var,
|
|
|
|
int coeff_shift) {
|
2016-08-15 20:27:19 +03:00
|
|
|
int i;
|
2016-08-19 01:10:22 +03:00
|
|
|
int32_t cost[8] = { 0 };
|
|
|
|
int partial[8][15] = { { 0 } };
|
2016-08-15 20:27:19 +03:00
|
|
|
int32_t best_cost = 0;
|
|
|
|
int best_dir = 0;
|
|
|
|
/* Instead of dividing by n between 2 and 8, we multiply by 3*5*7*8/n.
|
|
|
|
The output is then 840 times larger, but we don't care for finding
|
|
|
|
the max. */
|
2016-08-19 01:10:22 +03:00
|
|
|
static const int div_table[] = { 0, 840, 420, 280, 210, 168, 140, 120, 105 };
|
2016-08-15 20:27:19 +03:00
|
|
|
for (i = 0; i < 8; i++) {
|
|
|
|
int j;
|
|
|
|
for (j = 0; j < 8; j++) {
|
|
|
|
int x;
|
|
|
|
/* We subtract 128 here to reduce the maximum range of the squared
|
|
|
|
partial sums. */
|
2016-08-19 01:10:22 +03:00
|
|
|
x = (img[i * stride + j] >> coeff_shift) - 128;
|
2016-08-15 20:27:19 +03:00
|
|
|
partial[0][i + j] += x;
|
2016-08-19 01:10:22 +03:00
|
|
|
partial[1][i + j / 2] += x;
|
2016-08-15 20:27:19 +03:00
|
|
|
partial[2][i] += x;
|
2016-08-19 01:10:22 +03:00
|
|
|
partial[3][3 + i - j / 2] += x;
|
2016-08-15 20:27:19 +03:00
|
|
|
partial[4][7 + i - j] += x;
|
2016-08-19 01:10:22 +03:00
|
|
|
partial[5][3 - i / 2 + j] += x;
|
2016-08-15 20:27:19 +03:00
|
|
|
partial[6][j] += x;
|
2016-08-19 01:10:22 +03:00
|
|
|
partial[7][i / 2 + j] += x;
|
2016-08-15 20:27:19 +03:00
|
|
|
}
|
|
|
|
}
|
|
|
|
for (i = 0; i < 8; i++) {
|
2016-08-19 01:10:22 +03:00
|
|
|
cost[2] += partial[2][i] * partial[2][i];
|
|
|
|
cost[6] += partial[6][i] * partial[6][i];
|
2016-08-15 20:27:19 +03:00
|
|
|
}
|
|
|
|
cost[2] *= div_table[8];
|
|
|
|
cost[6] *= div_table[8];
|
|
|
|
for (i = 0; i < 7; i++) {
|
2016-08-19 01:10:22 +03:00
|
|
|
cost[0] += (partial[0][i] * partial[0][i] +
|
|
|
|
partial[0][14 - i] * partial[0][14 - i]) *
|
|
|
|
div_table[i + 1];
|
|
|
|
cost[4] += (partial[4][i] * partial[4][i] +
|
|
|
|
partial[4][14 - i] * partial[4][14 - i]) *
|
|
|
|
div_table[i + 1];
|
2016-08-15 20:27:19 +03:00
|
|
|
}
|
2016-08-19 01:10:22 +03:00
|
|
|
cost[0] += partial[0][7] * partial[0][7] * div_table[8];
|
|
|
|
cost[4] += partial[4][7] * partial[4][7] * div_table[8];
|
2016-08-15 20:27:19 +03:00
|
|
|
for (i = 1; i < 8; i += 2) {
|
|
|
|
int j;
|
|
|
|
for (j = 0; j < 4 + 1; j++) {
|
2016-08-19 01:10:22 +03:00
|
|
|
cost[i] += partial[i][3 + j] * partial[i][3 + j];
|
2016-08-15 20:27:19 +03:00
|
|
|
}
|
|
|
|
cost[i] *= div_table[8];
|
|
|
|
for (j = 0; j < 4 - 1; j++) {
|
2016-08-19 01:10:22 +03:00
|
|
|
cost[i] += (partial[i][j] * partial[i][j] +
|
|
|
|
partial[i][10 - j] * partial[i][10 - j]) *
|
|
|
|
div_table[2 * j + 2];
|
2016-08-15 20:27:19 +03:00
|
|
|
}
|
|
|
|
}
|
|
|
|
for (i = 0; i < 8; i++) {
|
|
|
|
if (cost[i] > best_cost) {
|
|
|
|
best_cost = cost[i];
|
|
|
|
best_dir = i;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
/* Difference between the optimal variance and the variance along the
|
|
|
|
orthogonal direction. Again, the sum(x^2) terms cancel out. */
|
|
|
|
*var = best_cost - cost[(best_dir + 4) & 7];
|
|
|
|
/* We'd normally divide by 840, but dividing by 1024 is close enough
|
|
|
|
for what we're going to do with this. */
|
|
|
|
*var >>= 10;
|
|
|
|
return best_dir;
|
|
|
|
}
|
|
|
|
|
|
|
|
#define OD_DERING_VERY_LARGE (30000)
|
2016-08-19 01:10:22 +03:00
|
|
|
#define OD_DERING_INBUF_SIZE \
|
|
|
|
((OD_BSIZE_MAX + 2 * OD_FILT_BORDER) * (OD_BSIZE_MAX + 2 * OD_FILT_BORDER))
|
2016-08-15 20:27:19 +03:00
|
|
|
|
|
|
|
/* Smooth in the direction detected. */
|
2016-09-16 18:06:50 +03:00
|
|
|
int od_filter_dering_direction_8x8_c(int16_t *y, int ystride, const int16_t *in,
|
|
|
|
int threshold, int dir) {
|
2016-08-15 20:27:19 +03:00
|
|
|
int i;
|
|
|
|
int j;
|
|
|
|
int k;
|
2016-09-01 21:35:16 +03:00
|
|
|
static const int taps[3] = { 3, 2, 1 };
|
2016-09-16 18:06:50 +03:00
|
|
|
int total_abs = 0;
|
2016-09-15 23:23:12 +03:00
|
|
|
for (i = 0; i < 8; i++) {
|
|
|
|
for (j = 0; j < 8; j++) {
|
2016-08-15 20:27:19 +03:00
|
|
|
int16_t sum;
|
|
|
|
int16_t xx;
|
|
|
|
int16_t yy;
|
2016-08-19 01:10:22 +03:00
|
|
|
xx = in[i * OD_FILT_BSTRIDE + j];
|
|
|
|
sum = 0;
|
2016-08-15 20:27:19 +03:00
|
|
|
for (k = 0; k < 3; k++) {
|
|
|
|
int16_t p0;
|
|
|
|
int16_t p1;
|
2016-08-19 01:10:22 +03:00
|
|
|
p0 = in[i * OD_FILT_BSTRIDE + j + OD_DIRECTION_OFFSETS_TABLE[dir][k]] -
|
|
|
|
xx;
|
|
|
|
p1 = in[i * OD_FILT_BSTRIDE + j - OD_DIRECTION_OFFSETS_TABLE[dir][k]] -
|
|
|
|
xx;
|
|
|
|
if (abs(p0) < threshold) sum += taps[k] * p0;
|
|
|
|
if (abs(p1) < threshold) sum += taps[k] * p1;
|
2016-08-15 20:27:19 +03:00
|
|
|
}
|
2016-09-16 18:06:50 +03:00
|
|
|
sum = (sum + 8) >> 4;
|
|
|
|
total_abs += abs(sum);
|
|
|
|
yy = xx + sum;
|
2016-08-19 01:10:22 +03:00
|
|
|
y[i * ystride + j] = yy;
|
2016-08-15 20:27:19 +03:00
|
|
|
}
|
|
|
|
}
|
2016-09-16 18:06:50 +03:00
|
|
|
return (total_abs + 8) >> 4;
|
2016-08-15 20:27:19 +03:00
|
|
|
}
|
|
|
|
|
2016-09-15 23:23:12 +03:00
|
|
|
/* Smooth in the direction detected. */
|
2016-09-16 18:06:50 +03:00
|
|
|
int od_filter_dering_direction_4x4_c(int16_t *y, int ystride, const int16_t *in,
|
|
|
|
int threshold, int dir) {
|
2016-09-15 23:23:12 +03:00
|
|
|
int i;
|
|
|
|
int j;
|
|
|
|
int k;
|
|
|
|
static const int taps[2] = { 4, 1 };
|
2016-09-16 18:06:50 +03:00
|
|
|
int total_abs = 0;
|
2016-09-15 23:23:12 +03:00
|
|
|
for (i = 0; i < 4; i++) {
|
|
|
|
for (j = 0; j < 4; j++) {
|
|
|
|
int16_t sum;
|
|
|
|
int16_t xx;
|
|
|
|
int16_t yy;
|
|
|
|
xx = in[i * OD_FILT_BSTRIDE + j];
|
|
|
|
sum = 0;
|
|
|
|
for (k = 0; k < 2; k++) {
|
|
|
|
int16_t p0;
|
|
|
|
int16_t p1;
|
|
|
|
p0 = in[i * OD_FILT_BSTRIDE + j + OD_DIRECTION_OFFSETS_TABLE[dir][k]] -
|
|
|
|
xx;
|
|
|
|
p1 = in[i * OD_FILT_BSTRIDE + j - OD_DIRECTION_OFFSETS_TABLE[dir][k]] -
|
|
|
|
xx;
|
|
|
|
if (abs(p0) < threshold) sum += taps[k] * p0;
|
|
|
|
if (abs(p1) < threshold) sum += taps[k] * p1;
|
|
|
|
}
|
2016-09-16 18:06:50 +03:00
|
|
|
sum = (sum + 8) >> 4;
|
|
|
|
total_abs += abs(sum);
|
|
|
|
yy = xx + sum;
|
2016-09-15 23:23:12 +03:00
|
|
|
y[i * ystride + j] = yy;
|
|
|
|
}
|
|
|
|
}
|
2016-09-16 18:06:50 +03:00
|
|
|
return (total_abs + 2) >> 2;
|
2016-08-15 20:27:19 +03:00
|
|
|
}
|
|
|
|
|
|
|
|
/* Smooth in the direction orthogonal to what was detected. */
|
2016-09-15 23:23:12 +03:00
|
|
|
void od_filter_dering_orthogonal_8x8_c(int16_t *y, int ystride,
|
2016-09-16 18:06:50 +03:00
|
|
|
const int16_t *in, int threshold,
|
|
|
|
int dir) {
|
2016-08-15 20:27:19 +03:00
|
|
|
int i;
|
|
|
|
int j;
|
|
|
|
int offset;
|
2016-08-19 01:10:22 +03:00
|
|
|
if (dir > 0 && dir < 4)
|
|
|
|
offset = OD_FILT_BSTRIDE;
|
|
|
|
else
|
|
|
|
offset = 1;
|
2016-09-15 23:23:12 +03:00
|
|
|
for (i = 0; i < 8; i++) {
|
|
|
|
for (j = 0; j < 8; j++) {
|
2016-08-15 20:27:19 +03:00
|
|
|
int16_t yy;
|
|
|
|
int16_t sum;
|
|
|
|
int16_t p;
|
2016-08-19 01:10:22 +03:00
|
|
|
yy = in[i * OD_FILT_BSTRIDE + j];
|
2016-08-15 20:27:19 +03:00
|
|
|
sum = 0;
|
2016-08-19 01:10:22 +03:00
|
|
|
p = in[i * OD_FILT_BSTRIDE + j + offset] - yy;
|
2016-09-16 18:06:50 +03:00
|
|
|
if (abs(p) < threshold) sum += p;
|
2016-08-19 01:10:22 +03:00
|
|
|
p = in[i * OD_FILT_BSTRIDE + j - offset] - yy;
|
2016-09-16 18:06:50 +03:00
|
|
|
if (abs(p) < threshold) sum += p;
|
2016-08-19 01:10:22 +03:00
|
|
|
p = in[i * OD_FILT_BSTRIDE + j + 2 * offset] - yy;
|
2016-09-16 18:06:50 +03:00
|
|
|
if (abs(p) < threshold) sum += p;
|
2016-08-19 01:10:22 +03:00
|
|
|
p = in[i * OD_FILT_BSTRIDE + j - 2 * offset] - yy;
|
2016-09-16 18:06:50 +03:00
|
|
|
if (abs(p) < threshold) sum += p;
|
2016-08-19 01:10:22 +03:00
|
|
|
y[i * ystride + j] = yy + ((3 * sum + 8) >> 4);
|
2016-08-15 20:27:19 +03:00
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2016-09-15 23:23:12 +03:00
|
|
|
/* Smooth in the direction orthogonal to what was detected. */
|
2016-08-15 20:27:19 +03:00
|
|
|
void od_filter_dering_orthogonal_4x4_c(int16_t *y, int ystride,
|
2016-09-16 18:06:50 +03:00
|
|
|
const int16_t *in, int threshold,
|
|
|
|
int dir) {
|
2016-09-15 23:23:12 +03:00
|
|
|
int i;
|
|
|
|
int j;
|
|
|
|
int offset;
|
|
|
|
if (dir > 0 && dir < 4)
|
|
|
|
offset = OD_FILT_BSTRIDE;
|
|
|
|
else
|
|
|
|
offset = 1;
|
|
|
|
for (i = 0; i < 4; i++) {
|
|
|
|
for (j = 0; j < 4; j++) {
|
|
|
|
int16_t yy;
|
|
|
|
int16_t sum;
|
|
|
|
int16_t p;
|
|
|
|
yy = in[i * OD_FILT_BSTRIDE + j];
|
|
|
|
sum = 0;
|
|
|
|
p = in[i * OD_FILT_BSTRIDE + j + offset] - yy;
|
2016-09-16 18:06:50 +03:00
|
|
|
if (abs(p) < threshold) sum += p;
|
2016-09-15 23:23:12 +03:00
|
|
|
p = in[i * OD_FILT_BSTRIDE + j - offset] - yy;
|
2016-09-16 18:06:50 +03:00
|
|
|
if (abs(p) < threshold) sum += p;
|
2016-09-15 23:23:12 +03:00
|
|
|
y[i * ystride + j] = yy + ((5 * sum + 8) >> 4);
|
|
|
|
}
|
|
|
|
}
|
2016-08-15 20:27:19 +03:00
|
|
|
}
|
|
|
|
|
|
|
|
/* This table approximates x^0.16 with the index being log2(x). It is clamped
|
|
|
|
to [-.5, 3]. The table is computed as:
|
|
|
|
round(256*min(3, max(.5, 1.08*(sqrt(2)*2.^([0:17]+8)/256/256).^.16))) */
|
|
|
|
static const int16_t OD_THRESH_TABLE_Q8[18] = {
|
2016-08-19 01:10:22 +03:00
|
|
|
128, 134, 150, 168, 188, 210, 234, 262, 292,
|
|
|
|
327, 365, 408, 455, 509, 569, 635, 710, 768,
|
2016-08-15 20:27:19 +03:00
|
|
|
};
|
|
|
|
|
2016-10-08 01:10:19 +03:00
|
|
|
/* Compute deringing filter threshold for an 8x8 block based on the
|
2016-08-15 20:27:19 +03:00
|
|
|
directional variance difference. A high variance difference means that we
|
|
|
|
have a highly directional pattern (e.g. a high contrast edge), so we can
|
|
|
|
apply more deringing. A low variance means that we either have a low
|
|
|
|
contrast edge, or a non-directional texture, so we want to be careful not
|
|
|
|
to blur. */
|
2016-10-08 01:10:19 +03:00
|
|
|
static INLINE int od_adjust_thresh(int threshold, int32_t var) {
|
|
|
|
int v1;
|
|
|
|
/* We use the variance of 8x8 blocks to adjust the threshold. */
|
|
|
|
v1 = OD_MINI(32767, var >> 6);
|
|
|
|
return (threshold * OD_THRESH_TABLE_Q8[OD_ILOG(v1)] + 128) >> 8;
|
2016-08-15 20:27:19 +03:00
|
|
|
}
|
|
|
|
|
2016-10-12 00:47:36 +03:00
|
|
|
static INLINE void copy_8x8_16bit(int16_t *dst, int dstride, int16_t *src, int sstride) {
|
|
|
|
int i, j;
|
|
|
|
for (i = 0; i < 8; i++)
|
|
|
|
for (j = 0; j < 8; j++)
|
|
|
|
dst[i * dstride + j] = src[i * sstride + j];
|
|
|
|
}
|
|
|
|
|
|
|
|
static INLINE void copy_4x4_16bit(int16_t *dst, int dstride, int16_t *src, int sstride) {
|
|
|
|
int i, j;
|
|
|
|
for (i = 0; i < 4; i++)
|
|
|
|
for (j = 0; j < 4; j++)
|
|
|
|
dst[i * dstride + j] = src[i * sstride + j];
|
|
|
|
}
|
|
|
|
|
|
|
|
/* TODO: Optimize this function for SSE. */
|
2016-10-13 01:09:48 +03:00
|
|
|
void copy_blocks_16bit(int16_t *dst, int dstride, int16_t *src,
|
2016-10-12 00:47:36 +03:00
|
|
|
unsigned char (*bskip)[2], int dering_count, int bsize)
|
|
|
|
{
|
|
|
|
int bi, bx, by;
|
|
|
|
if (bsize == 3) {
|
|
|
|
for (bi = 0; bi < dering_count; bi++) {
|
|
|
|
by = bskip[bi][0];
|
|
|
|
bx = bskip[bi][1];
|
|
|
|
copy_8x8_16bit(&dst[(by << 3) * dstride + (bx << 3)],
|
|
|
|
dstride,
|
2016-10-13 01:09:48 +03:00
|
|
|
&src[bi << 2*bsize], 1 << bsize);
|
2016-10-12 00:47:36 +03:00
|
|
|
}
|
|
|
|
} else {
|
|
|
|
for (bi = 0; bi < dering_count; bi++) {
|
|
|
|
by = bskip[bi][0];
|
|
|
|
bx = bskip[bi][1];
|
|
|
|
copy_4x4_16bit(&dst[(by << 2) * dstride + (bx << 2)],
|
|
|
|
dstride,
|
2016-10-13 01:09:48 +03:00
|
|
|
&src[bi << 2*bsize], 1 << bsize);
|
2016-10-12 00:47:36 +03:00
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2016-10-13 01:09:48 +03:00
|
|
|
void od_dering(int16_t *y, const od_dering_in *x, int xstride,
|
2016-10-07 00:27:34 +03:00
|
|
|
int nhb, int nvb, int sbx, int sby, int nhsb, int nvsb, int xdec,
|
2016-11-02 18:22:02 +03:00
|
|
|
int dir[OD_DERING_NBLOCKS][OD_DERING_NBLOCKS], int pli,
|
2016-10-11 23:53:59 +03:00
|
|
|
unsigned char (*bskip)[2], int dering_count, int threshold,
|
2016-08-23 09:45:43 +03:00
|
|
|
int coeff_shift) {
|
2016-08-15 20:27:19 +03:00
|
|
|
int i;
|
|
|
|
int j;
|
2016-10-11 23:53:59 +03:00
|
|
|
int bi;
|
2016-08-15 20:27:19 +03:00
|
|
|
int bx;
|
|
|
|
int by;
|
|
|
|
int16_t inbuf[OD_DERING_INBUF_SIZE];
|
|
|
|
int16_t *in;
|
2016-11-02 18:22:02 +03:00
|
|
|
int bsize;
|
2016-10-08 01:10:19 +03:00
|
|
|
int filter2_thresh[OD_DERING_NBLOCKS][OD_DERING_NBLOCKS];
|
2016-10-07 00:27:34 +03:00
|
|
|
od_filter_dering_direction_func filter_dering_direction[OD_DERINGSIZES] = {
|
2016-11-02 18:22:02 +03:00
|
|
|
od_filter_dering_direction_4x4, od_filter_dering_direction_8x8
|
2016-10-07 00:27:34 +03:00
|
|
|
};
|
|
|
|
od_filter_dering_orthogonal_func filter_dering_orthogonal[OD_DERINGSIZES] = {
|
2016-11-02 18:22:02 +03:00
|
|
|
od_filter_dering_orthogonal_4x4, od_filter_dering_orthogonal_8x8
|
2016-10-07 00:27:34 +03:00
|
|
|
};
|
2016-11-02 18:22:02 +03:00
|
|
|
bsize = 3 - xdec;
|
2016-08-19 01:10:22 +03:00
|
|
|
in = inbuf + OD_FILT_BORDER * OD_FILT_BSTRIDE + OD_FILT_BORDER;
|
2016-08-15 20:27:19 +03:00
|
|
|
/* We avoid filtering the pixels for which some of the pixels to average
|
|
|
|
are outside the frame. We could change the filter instead, but it would
|
|
|
|
add special cases for any future vectorization. */
|
|
|
|
for (i = 0; i < OD_DERING_INBUF_SIZE; i++) inbuf[i] = OD_DERING_VERY_LARGE;
|
2016-08-19 01:10:22 +03:00
|
|
|
for (i = -OD_FILT_BORDER * (sby != 0);
|
2016-11-02 18:22:02 +03:00
|
|
|
i < (nvb << bsize) + OD_FILT_BORDER * (sby != nvsb - 1); i++) {
|
2016-08-19 01:10:22 +03:00
|
|
|
for (j = -OD_FILT_BORDER * (sbx != 0);
|
2016-11-02 18:22:02 +03:00
|
|
|
j < (nhb << bsize) + OD_FILT_BORDER * (sbx != nhsb - 1); j++) {
|
2016-08-19 01:10:22 +03:00
|
|
|
in[i * OD_FILT_BSTRIDE + j] = x[i * xstride + j];
|
2016-08-15 20:27:19 +03:00
|
|
|
}
|
|
|
|
}
|
|
|
|
if (pli == 0) {
|
2016-10-11 23:53:59 +03:00
|
|
|
for (bi = 0; bi < dering_count; bi++) {
|
2016-10-13 00:45:07 +03:00
|
|
|
int32_t var;
|
2016-10-11 23:53:59 +03:00
|
|
|
by = bskip[bi][0];
|
|
|
|
bx = bskip[bi][1];
|
|
|
|
dir[by][bx] = od_dir_find8(&x[8 * by * xstride + 8 * bx], xstride,
|
2016-10-13 00:45:07 +03:00
|
|
|
&var, coeff_shift);
|
2016-10-11 23:53:59 +03:00
|
|
|
/* Deringing orthogonal to the direction uses a tighter threshold
|
|
|
|
because we want to be conservative. We've presumably already
|
|
|
|
achieved some deringing, so the amount of change is expected
|
|
|
|
to be low. Also, since we might be filtering across an edge, we
|
|
|
|
want to make sure not to blur it. That being said, we might want
|
|
|
|
to be a little bit more aggressive on pure horizontal/vertical
|
|
|
|
since the ringing there tends to be directional, so it doesn't
|
|
|
|
get removed by the directional filtering. */
|
|
|
|
filter2_thresh[by][bx] = (filter_dering_direction[bsize - OD_LOG_BSIZE0])(
|
2016-10-13 01:09:48 +03:00
|
|
|
&y[bi << 2*bsize], 1 << bsize,
|
2016-10-11 23:53:59 +03:00
|
|
|
&in[(by * OD_FILT_BSTRIDE << bsize) + (bx << bsize)],
|
2016-10-13 00:45:07 +03:00
|
|
|
od_adjust_thresh(threshold, var), dir[by][bx]);
|
2016-08-15 20:27:19 +03:00
|
|
|
}
|
2016-08-19 01:10:22 +03:00
|
|
|
} else {
|
2016-10-11 23:53:59 +03:00
|
|
|
for (bi = 0; bi < dering_count; bi++) {
|
|
|
|
by = bskip[bi][0];
|
|
|
|
bx = bskip[bi][1];
|
|
|
|
filter2_thresh[by][bx] = (filter_dering_direction[bsize - OD_LOG_BSIZE0])(
|
2016-10-13 01:09:48 +03:00
|
|
|
&y[bi << 2*bsize], 1 << bsize,
|
2016-10-11 23:53:59 +03:00
|
|
|
&in[(by * OD_FILT_BSTRIDE << bsize) + (bx << bsize)], threshold,
|
|
|
|
dir[by][bx]);
|
2016-08-15 20:27:19 +03:00
|
|
|
}
|
|
|
|
}
|
2016-10-13 01:09:48 +03:00
|
|
|
copy_blocks_16bit(in, OD_FILT_BSTRIDE, y, bskip, dering_count,
|
2016-10-12 00:47:36 +03:00
|
|
|
bsize);
|
2016-10-11 23:53:59 +03:00
|
|
|
for (bi = 0; bi < dering_count; bi++) {
|
|
|
|
by = bskip[bi][0];
|
|
|
|
bx = bskip[bi][1];
|
|
|
|
if (filter2_thresh[by][bx] == 0) continue;
|
|
|
|
(filter_dering_orthogonal[bsize - OD_LOG_BSIZE0])(
|
2016-10-13 01:09:48 +03:00
|
|
|
&y[bi << 2*bsize], 1 << bsize,
|
2016-10-11 23:53:59 +03:00
|
|
|
&in[(by * OD_FILT_BSTRIDE << bsize) + (bx << bsize)], filter2_thresh[by][bx],
|
|
|
|
dir[by][bx]);
|
2016-08-15 20:27:19 +03:00
|
|
|
}
|
|
|
|
}
|