mirror of
https://git.ffmpeg.org/ffmpeg.git
synced 2025-01-01 04:12:14 +00:00
avfilter: add xmedian filter
This commit is contained in:
parent
87a54e150e
commit
8ef163ba9e
@ -28,6 +28,7 @@ version <next>:
|
||||
- asoftclip filter
|
||||
- Support decoding of HEVC 4:4:4 content in vdpau
|
||||
- colorhold filter
|
||||
- xmedian filter
|
||||
|
||||
|
||||
version 4.1:
|
||||
|
@ -18418,6 +18418,20 @@ Set the scaling dimension: @code{2} for @code{2xBR}, @code{3} for
|
||||
Default is @code{3}.
|
||||
@end table
|
||||
|
||||
@section xmedian
|
||||
Pick median pixels from several input videos.
|
||||
|
||||
The filter accept the following options:
|
||||
|
||||
@table @option
|
||||
@item nb_inputs
|
||||
Set number of inputs. This must be odd number.
|
||||
Default is 3. Allowed range is from 3 to 255.
|
||||
|
||||
@item planes
|
||||
Set which planes to filter. Default value is @code{15}, by which all planes are processed.
|
||||
@end table
|
||||
|
||||
@section xstack
|
||||
Stack video inputs into custom layout.
|
||||
|
||||
|
@ -421,6 +421,7 @@ OBJS-$(CONFIG_W3FDIF_FILTER) += vf_w3fdif.o
|
||||
OBJS-$(CONFIG_WAVEFORM_FILTER) += vf_waveform.o
|
||||
OBJS-$(CONFIG_WEAVE_FILTER) += vf_weave.o
|
||||
OBJS-$(CONFIG_XBR_FILTER) += vf_xbr.o
|
||||
OBJS-$(CONFIG_XMEDIAN_FILTER) += vf_xmedian.o framesync.o
|
||||
OBJS-$(CONFIG_XSTACK_FILTER) += vf_stack.o framesync.o
|
||||
OBJS-$(CONFIG_YADIF_FILTER) += vf_yadif.o yadif_common.o
|
||||
OBJS-$(CONFIG_YADIF_CUDA_FILTER) += vf_yadif_cuda.o vf_yadif_cuda.ptx.o \
|
||||
|
@ -400,6 +400,7 @@ extern AVFilter ff_vf_w3fdif;
|
||||
extern AVFilter ff_vf_waveform;
|
||||
extern AVFilter ff_vf_weave;
|
||||
extern AVFilter ff_vf_xbr;
|
||||
extern AVFilter ff_vf_xmedian;
|
||||
extern AVFilter ff_vf_xstack;
|
||||
extern AVFilter ff_vf_yadif;
|
||||
extern AVFilter ff_vf_yadif_cuda;
|
||||
|
@ -30,7 +30,7 @@
|
||||
#include "libavutil/version.h"
|
||||
|
||||
#define LIBAVFILTER_VERSION_MAJOR 7
|
||||
#define LIBAVFILTER_VERSION_MINOR 51
|
||||
#define LIBAVFILTER_VERSION_MINOR 52
|
||||
#define LIBAVFILTER_VERSION_MICRO 100
|
||||
|
||||
#define LIBAVFILTER_VERSION_INT AV_VERSION_INT(LIBAVFILTER_VERSION_MAJOR, \
|
||||
|
347
libavfilter/vf_xmedian.c
Normal file
347
libavfilter/vf_xmedian.c
Normal file
@ -0,0 +1,347 @@
|
||||
/*
|
||||
* Copyright (c) 2019 Paul B Mahol
|
||||
*
|
||||
* This file is part of FFmpeg.
|
||||
*
|
||||
* FFmpeg is free software; you can redistribute it and/or
|
||||
* modify it under the terms of the GNU Lesser General Public
|
||||
* License as published by the Free Software Foundation; either
|
||||
* version 2.1 of the License, or (at your option) any later version.
|
||||
*
|
||||
* FFmpeg 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
|
||||
* Lesser General Public License for more details.
|
||||
*
|
||||
* You should have received a copy of the GNU Lesser General Public
|
||||
* License along with FFmpeg; if not, write to the Free Software
|
||||
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
|
||||
*/
|
||||
|
||||
#include "libavutil/avstring.h"
|
||||
#include "libavutil/imgutils.h"
|
||||
#include "libavutil/intreadwrite.h"
|
||||
#include "libavutil/opt.h"
|
||||
#include "libavutil/pixdesc.h"
|
||||
#include "libavutil/qsort.h"
|
||||
|
||||
#include "avfilter.h"
|
||||
#include "formats.h"
|
||||
#include "internal.h"
|
||||
#include "framesync.h"
|
||||
#include "video.h"
|
||||
|
||||
typedef struct XMedianContext {
|
||||
const AVClass *class;
|
||||
const AVPixFmtDescriptor *desc;
|
||||
int nb_inputs;
|
||||
int planes;
|
||||
|
||||
int radius;
|
||||
int depth;
|
||||
int max;
|
||||
int nb_planes;
|
||||
int linesize[4];
|
||||
int width[4];
|
||||
int height[4];
|
||||
|
||||
AVFrame **frames;
|
||||
FFFrameSync fs;
|
||||
|
||||
int (*median_frames)(AVFilterContext *ctx, void *arg, int jobnr, int nb_jobs);
|
||||
} XMedianContext;
|
||||
|
||||
static int query_formats(AVFilterContext *ctx)
|
||||
{
|
||||
static const enum AVPixelFormat pixel_fmts[] = {
|
||||
AV_PIX_FMT_GRAY8,
|
||||
AV_PIX_FMT_GRAY9,
|
||||
AV_PIX_FMT_GRAY10,
|
||||
AV_PIX_FMT_GRAY12,
|
||||
AV_PIX_FMT_GRAY14,
|
||||
AV_PIX_FMT_GRAY16,
|
||||
AV_PIX_FMT_YUV410P, AV_PIX_FMT_YUV411P,
|
||||
AV_PIX_FMT_YUV420P, AV_PIX_FMT_YUV422P,
|
||||
AV_PIX_FMT_YUV440P, AV_PIX_FMT_YUV444P,
|
||||
AV_PIX_FMT_YUVJ420P, AV_PIX_FMT_YUVJ422P,
|
||||
AV_PIX_FMT_YUVJ440P, AV_PIX_FMT_YUVJ444P,
|
||||
AV_PIX_FMT_YUVJ411P,
|
||||
AV_PIX_FMT_YUV420P9, AV_PIX_FMT_YUV422P9, AV_PIX_FMT_YUV444P9,
|
||||
AV_PIX_FMT_YUV420P10, AV_PIX_FMT_YUV422P10, AV_PIX_FMT_YUV444P10,
|
||||
AV_PIX_FMT_YUV440P10,
|
||||
AV_PIX_FMT_YUV444P12, AV_PIX_FMT_YUV422P12, AV_PIX_FMT_YUV420P12,
|
||||
AV_PIX_FMT_YUV440P12,
|
||||
AV_PIX_FMT_YUV444P14, AV_PIX_FMT_YUV422P14, AV_PIX_FMT_YUV420P14,
|
||||
AV_PIX_FMT_YUV420P16, AV_PIX_FMT_YUV422P16, AV_PIX_FMT_YUV444P16,
|
||||
AV_PIX_FMT_GBRP, AV_PIX_FMT_GBRP9, AV_PIX_FMT_GBRP10,
|
||||
AV_PIX_FMT_GBRP12, AV_PIX_FMT_GBRP14, AV_PIX_FMT_GBRP16,
|
||||
AV_PIX_FMT_NONE
|
||||
};
|
||||
AVFilterFormats *formats = ff_make_format_list(pixel_fmts);
|
||||
if (!formats)
|
||||
return AVERROR(ENOMEM);
|
||||
return ff_set_common_formats(ctx, formats);
|
||||
}
|
||||
|
||||
static av_cold int init(AVFilterContext *ctx)
|
||||
{
|
||||
XMedianContext *s = ctx->priv;
|
||||
int ret;
|
||||
|
||||
if (!(s->nb_inputs & 1))
|
||||
av_log(s, AV_LOG_WARNING, "nb_intputs: %d is not odd number.\n", s->nb_inputs);
|
||||
|
||||
s->nb_inputs = s->nb_inputs | 1;
|
||||
s->radius = s->nb_inputs / 2;
|
||||
s->frames = av_calloc(s->nb_inputs, sizeof(*s->frames));
|
||||
if (!s->frames)
|
||||
return AVERROR(ENOMEM);
|
||||
|
||||
for (int i = 0; i < s->nb_inputs; i++) {
|
||||
AVFilterPad pad = { 0 };
|
||||
|
||||
pad.type = AVMEDIA_TYPE_VIDEO;
|
||||
pad.name = av_asprintf("input%d", i);
|
||||
if (!pad.name)
|
||||
return AVERROR(ENOMEM);
|
||||
|
||||
if ((ret = ff_insert_inpad(ctx, i, &pad)) < 0) {
|
||||
av_freep(&pad.name);
|
||||
return ret;
|
||||
}
|
||||
}
|
||||
|
||||
return 0;
|
||||
}
|
||||
|
||||
typedef struct ThreadData {
|
||||
AVFrame **in, *out;
|
||||
} ThreadData;
|
||||
|
||||
static int comparei(const void *p1, const void *p2)
|
||||
{
|
||||
int left = *(const int *)p1;
|
||||
int right = *(const int *)p2;
|
||||
return FFDIFFSIGN(left, right);
|
||||
}
|
||||
|
||||
static int median_frames16(AVFilterContext *ctx, void *arg, int jobnr, int nb_jobs)
|
||||
{
|
||||
XMedianContext *s = ctx->priv;
|
||||
ThreadData *td = arg;
|
||||
AVFrame **in = td->in;
|
||||
AVFrame *out = td->out;
|
||||
const int nb_inputs = s->nb_inputs;
|
||||
const int radius = s->radius;
|
||||
int values[256];
|
||||
|
||||
for (int p = 0; p < s->nb_planes; p++) {
|
||||
const int slice_start = (s->height[p] * jobnr) / nb_jobs;
|
||||
const int slice_end = (s->height[p] * (jobnr+1)) / nb_jobs;
|
||||
uint16_t *dst = (uint16_t *)(out->data[p] + slice_start * out->linesize[p]);
|
||||
|
||||
if (!((1 << p) & s->planes)) {
|
||||
av_image_copy_plane((uint8_t *)dst, out->linesize[p],
|
||||
in[0]->data[p] + slice_start * in[radius]->linesize[p],
|
||||
in[0]->linesize[p],
|
||||
s->linesize[p], slice_end - slice_start);
|
||||
continue;
|
||||
}
|
||||
|
||||
for (int y = slice_start; y < slice_end; y++) {
|
||||
for (int x = 0; x < s->width[p]; x++) {
|
||||
for (int i = 0; i < nb_inputs; i++) {
|
||||
const uint16_t *src = (const uint16_t *)(in[i]->data[p] + y * in[i]->linesize[p]);
|
||||
values[i] = src[x];
|
||||
}
|
||||
|
||||
AV_QSORT(values, nb_inputs, int, comparei);
|
||||
dst[x] = values[radius];
|
||||
}
|
||||
|
||||
dst += out->linesize[p] / 2;
|
||||
}
|
||||
}
|
||||
|
||||
return 0;
|
||||
}
|
||||
|
||||
static int median_frames8(AVFilterContext *ctx, void *arg, int jobnr, int nb_jobs)
|
||||
{
|
||||
XMedianContext *s = ctx->priv;
|
||||
ThreadData *td = arg;
|
||||
AVFrame **in = td->in;
|
||||
AVFrame *out = td->out;
|
||||
const int nb_inputs = s->nb_inputs;
|
||||
const int radius = s->radius;
|
||||
int values[256];
|
||||
|
||||
for (int p = 0; p < s->nb_planes; p++) {
|
||||
const int slice_start = (s->height[p] * jobnr) / nb_jobs;
|
||||
const int slice_end = (s->height[p] * (jobnr+1)) / nb_jobs;
|
||||
uint8_t *dst = out->data[p] + slice_start * out->linesize[p];
|
||||
|
||||
if (!((1 << p) & s->planes)) {
|
||||
av_image_copy_plane(dst, out->linesize[p],
|
||||
in[0]->data[p] + slice_start * in[0]->linesize[p],
|
||||
in[0]->linesize[p],
|
||||
s->linesize[p], slice_end - slice_start);
|
||||
continue;
|
||||
}
|
||||
|
||||
for (int y = slice_start; y < slice_end; y++) {
|
||||
for (int x = 0; x < s->width[p]; x++) {
|
||||
for (int i = 0; i < nb_inputs; i++)
|
||||
values[i] = in[i]->data[p][y * in[i]->linesize[p] + x];
|
||||
|
||||
AV_QSORT(values, nb_inputs, int, comparei);
|
||||
dst[x] = values[radius];
|
||||
}
|
||||
|
||||
dst += out->linesize[p];
|
||||
}
|
||||
}
|
||||
|
||||
return 0;
|
||||
}
|
||||
|
||||
static int process_frame(FFFrameSync *fs)
|
||||
{
|
||||
AVFilterContext *ctx = fs->parent;
|
||||
AVFilterLink *outlink = ctx->outputs[0];
|
||||
XMedianContext *s = fs->opaque;
|
||||
AVFrame **in = s->frames;
|
||||
AVFrame *out;
|
||||
ThreadData td;
|
||||
int i, ret;
|
||||
|
||||
for (i = 0; i < s->nb_inputs; i++) {
|
||||
if ((ret = ff_framesync_get_frame(&s->fs, i, &in[i], 0)) < 0)
|
||||
return ret;
|
||||
}
|
||||
|
||||
out = ff_get_video_buffer(outlink, outlink->w, outlink->h);
|
||||
if (!out)
|
||||
return AVERROR(ENOMEM);
|
||||
out->pts = av_rescale_q(s->fs.pts, s->fs.time_base, outlink->time_base);
|
||||
|
||||
td.in = in;
|
||||
td.out = out;
|
||||
ctx->internal->execute(ctx, s->median_frames, &td, NULL, FFMIN(s->height[1], ff_filter_get_nb_threads(ctx)));
|
||||
|
||||
return ff_filter_frame(outlink, out);
|
||||
}
|
||||
|
||||
static int config_output(AVFilterLink *outlink)
|
||||
{
|
||||
AVFilterContext *ctx = outlink->src;
|
||||
XMedianContext *s = ctx->priv;
|
||||
AVRational frame_rate = ctx->inputs[0]->frame_rate;
|
||||
AVFilterLink *inlink = ctx->inputs[0];
|
||||
int height = ctx->inputs[0]->h;
|
||||
int width = ctx->inputs[0]->w;
|
||||
FFFrameSyncIn *in;
|
||||
int i, ret;
|
||||
|
||||
for (int i = 1; i < s->nb_inputs; i++) {
|
||||
if (ctx->inputs[i]->h != height || ctx->inputs[i]->w != width) {
|
||||
av_log(ctx, AV_LOG_ERROR, "Input %d size (%dx%d) does not match input %d size (%dx%d).\n", i, ctx->inputs[i]->w, ctx->inputs[i]->h, 0, width, height);
|
||||
return AVERROR(EINVAL);
|
||||
}
|
||||
}
|
||||
|
||||
s->desc = av_pix_fmt_desc_get(outlink->format);
|
||||
if (!s->desc)
|
||||
return AVERROR_BUG;
|
||||
s->nb_planes = av_pix_fmt_count_planes(outlink->format);
|
||||
s->depth = s->desc->comp[0].depth;
|
||||
s->max = (1 << s->depth) - 1;
|
||||
|
||||
if (s->depth <= 8)
|
||||
s->median_frames = median_frames8;
|
||||
else
|
||||
s->median_frames = median_frames16;
|
||||
|
||||
if ((ret = av_image_fill_linesizes(s->linesize, inlink->format, inlink->w)) < 0)
|
||||
return ret;
|
||||
|
||||
s->width[1] = s->width[2] = AV_CEIL_RSHIFT(inlink->w, s->desc->log2_chroma_w);
|
||||
s->width[0] = s->width[3] = inlink->w;
|
||||
s->height[1] = s->height[2] = AV_CEIL_RSHIFT(inlink->h, s->desc->log2_chroma_h);
|
||||
s->height[0] = s->height[3] = inlink->h;
|
||||
|
||||
outlink->w = width;
|
||||
outlink->h = height;
|
||||
outlink->frame_rate = frame_rate;
|
||||
|
||||
if ((ret = ff_framesync_init(&s->fs, ctx, s->nb_inputs)) < 0)
|
||||
return ret;
|
||||
|
||||
in = s->fs.in;
|
||||
s->fs.opaque = s;
|
||||
s->fs.on_event = process_frame;
|
||||
|
||||
for (i = 0; i < s->nb_inputs; i++) {
|
||||
AVFilterLink *inlink = ctx->inputs[i];
|
||||
|
||||
in[i].time_base = inlink->time_base;
|
||||
in[i].sync = 1;
|
||||
in[i].before = EXT_STOP;
|
||||
in[i].after = EXT_STOP;
|
||||
}
|
||||
|
||||
ret = ff_framesync_configure(&s->fs);
|
||||
outlink->time_base = s->fs.time_base;
|
||||
|
||||
return ret;
|
||||
}
|
||||
|
||||
static av_cold void uninit(AVFilterContext *ctx)
|
||||
{
|
||||
XMedianContext *s = ctx->priv;
|
||||
|
||||
ff_framesync_uninit(&s->fs);
|
||||
av_freep(&s->frames);
|
||||
|
||||
for (int i = 0; i < ctx->nb_inputs; i++)
|
||||
av_freep(&ctx->input_pads[i].name);
|
||||
}
|
||||
|
||||
static int activate(AVFilterContext *ctx)
|
||||
{
|
||||
XMedianContext *s = ctx->priv;
|
||||
return ff_framesync_activate(&s->fs);
|
||||
}
|
||||
|
||||
#define OFFSET(x) offsetof(XMedianContext, x)
|
||||
#define FLAGS AV_OPT_FLAG_VIDEO_PARAM | AV_OPT_FLAG_FILTERING_PARAM
|
||||
|
||||
static const AVOption xmedian_options[] = {
|
||||
{ "nb_inputs", "set number of inputs", OFFSET(nb_inputs), AV_OPT_TYPE_INT, {.i64=3}, 3, 255, .flags = FLAGS },
|
||||
{ "planes", "set planes to filter", OFFSET(planes), AV_OPT_TYPE_INT, {.i64=15}, 0, 15, .flags = FLAGS },
|
||||
{ NULL },
|
||||
};
|
||||
|
||||
static const AVFilterPad outputs[] = {
|
||||
{
|
||||
.name = "default",
|
||||
.type = AVMEDIA_TYPE_VIDEO,
|
||||
.config_props = config_output,
|
||||
},
|
||||
{ NULL }
|
||||
};
|
||||
|
||||
AVFILTER_DEFINE_CLASS(xmedian);
|
||||
|
||||
AVFilter ff_vf_xmedian = {
|
||||
.name = "xmedian",
|
||||
.description = NULL_IF_CONFIG_SMALL("Pick median pixels from several video inputs."),
|
||||
.priv_size = sizeof(XMedianContext),
|
||||
.priv_class = &xmedian_class,
|
||||
.query_formats = query_formats,
|
||||
.outputs = outputs,
|
||||
.init = init,
|
||||
.uninit = uninit,
|
||||
.activate = activate,
|
||||
.flags = AVFILTER_FLAG_DYNAMIC_INPUTS | AVFILTER_FLAG_SLICE_THREADS,
|
||||
};
|
Loading…
Reference in New Issue
Block a user