[FFmpeg-devel] [PATCH] lavfi: port ow from libmpcodecs.

Stefano Sabatini stefasab at gmail.com
Tue May 14 22:59:44 CEST 2013


On date Friday 2013-05-10 22:10:15 +0200, Clément Bœsch encoded:
> TODO: Changelog & minor bump
> 
> ---
> Feel free to bikeshed for a new name.
> 
> mp=ow will be dropped in the same time.
> ---
>  doc/filters.texi         |  26 ++++
>  libavfilter/Makefile     |   1 +
>  libavfilter/allfilters.c |   1 +
>  libavfilter/vf_ow.c      | 344 +++++++++++++++++++++++++++++++++++++++++++++++
>  4 files changed, 372 insertions(+)
>  create mode 100644 libavfilter/vf_ow.c
> 
> diff --git a/doc/filters.texi b/doc/filters.texi
> index fe38b7f..8b9fb4a 100644
> --- a/doc/filters.texi
> +++ b/doc/filters.texi
> @@ -5103,6 +5103,32 @@ testsrc=s=100x100, split=4 [in0][in1][in2][in3];
>  
>  @end itemize
>  
> + at section ow
> +
> +Overcomplete Wavelet denoiser.

Maybe tell when/if this filter is useful compared to other similar
tools.

> +
> +The filter accepts the following options:
> +
> + at table @option

> + at item depth
> +Set depth.

Depth of the sea?

> +
> +Larger depth values will denoise lower frequency components more, but
> +slow down filtering

Missing final dot.

> +
> +Default is @code{8}.
> +
> + at item luma_strength
> +Set luma strength.
> +
> +Default is @code{1.0}
> +
> + at item chroma_strength
> +Set chroma strength.
> +
> +Default is @code{1.0}

Also some notes about the strenght is about would be nice.

> + at end table
> +
>  @section pad
>  
>  Add paddings to the input image, and place the original input at the
> diff --git a/libavfilter/Makefile b/libavfilter/Makefile
> index 73a24dd..f6daed1 100644
> --- a/libavfilter/Makefile
> +++ b/libavfilter/Makefile
> @@ -154,6 +154,7 @@ OBJS-$(CONFIG_NULL_FILTER)                   += vf_null.o
>  OBJS-$(CONFIG_OCV_FILTER)                    += vf_libopencv.o
>  OBJS-$(CONFIG_OPENCL)                        += deshake_opencl.o unsharp_opencl.o
>  OBJS-$(CONFIG_OVERLAY_FILTER)                += vf_overlay.o
> +OBJS-$(CONFIG_OW_FILTER)                     += vf_ow.o
>  OBJS-$(CONFIG_PAD_FILTER)                    += vf_pad.o
>  OBJS-$(CONFIG_PERMS_FILTER)                  += f_perms.o
>  OBJS-$(CONFIG_PIXDESCTEST_FILTER)            += vf_pixdesctest.o
> diff --git a/libavfilter/allfilters.c b/libavfilter/allfilters.c
> index b8f273d..8986547 100644
> --- a/libavfilter/allfilters.c
> +++ b/libavfilter/allfilters.c
> @@ -151,6 +151,7 @@ void avfilter_register_all(void)
>      REGISTER_FILTER(NULL,           null,           vf);
>      REGISTER_FILTER(OCV,            ocv,            vf);
>      REGISTER_FILTER(OVERLAY,        overlay,        vf);
> +    REGISTER_FILTER(OW,             ow,             vf);
>      REGISTER_FILTER(PAD,            pad,            vf);
>      REGISTER_FILTER(PERMS,          perms,          vf);
>      REGISTER_FILTER(PIXDESCTEST,    pixdesctest,    vf);
> diff --git a/libavfilter/vf_ow.c b/libavfilter/vf_ow.c
> new file mode 100644
> index 0000000..97c386a
> --- /dev/null
> +++ b/libavfilter/vf_ow.c
> @@ -0,0 +1,344 @@

> +/*
> + * Copyright (C) 2007 Michael Niedermayer <michaelni at gmx.at>
> + * Copyright (c) 2013 Clément Bœsch
> + *
> + * 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

Keep original licensing or ask the author (Michael seems the only one
who did significant changes, things may be different on awesome
lawyers planet) to relicense it.

> + */
> +
> +/**
> + * @todo try to change to int
> + * @todo try lifting based implementation
> + * @todo optimize optimize optimize

> + * @todo hard tresholding

thresholding

> + * @todo use QP to decide filter strength
> + * @todo wavelet normalization / least squares optimal signal vs. noise thresholds
> + */
> +
> +#include "libavutil/imgutils.h"
> +#include "libavutil/opt.h"
> +#include "libavutil/pixdesc.h"
> +#include "avfilter.h"
> +#include "internal.h"
> +
> +typedef struct {
> +    const AVClass *class;
> +    double luma_strength;
> +    double chroma_strength;
> +    int depth;
> +    float *plane[16+1][4];
> +    int linesize;
> +    int hsub, vsub;
> +} OWContext;
> +
> +#define OFFSET(x) offsetof(OWContext, x)
> +#define FLAGS AV_OPT_FLAG_FILTERING_PARAM|AV_OPT_FLAG_VIDEO_PARAM
> +static const AVOption ow_options[] = {
> +    { "depth",           "set depth",           OFFSET(depth),           AV_OPT_TYPE_INT,    {.i64 = 8},    8,   16, FLAGS },
> +    { "luma_strength",   "set luma strength",   OFFSET(luma_strength),   AV_OPT_TYPE_DOUBLE, {.dbl = 1.0 }, 0, 1000, FLAGS },
> +    { "ls",              "set luma strength",   OFFSET(luma_strength),   AV_OPT_TYPE_DOUBLE, {.dbl = 1.0 }, 0, 1000, FLAGS },
> +    { "chroma_strength", "set chroma strength", OFFSET(chroma_strength), AV_OPT_TYPE_DOUBLE, {.dbl = 1.0 }, 0, 1000, FLAGS },
> +    { "cs",              "set chroma strength", OFFSET(chroma_strength), AV_OPT_TYPE_DOUBLE, {.dbl = 1.0 }, 0, 1000, FLAGS },
> +    { NULL }
> +};
> +
> +AVFILTER_DEFINE_CLASS(ow);
> +
> +DECLARE_ALIGNED(8, static const uint8_t, dither)[8][8] = {
> +    {  0,  48,  12,  60,   3,  51,  15,  63 },
> +    { 32,  16,  44,  28,  35,  19,  47,  31 },
> +    {  8,  56,   4,  52,  11,  59,   7,  55 },
> +    { 40,  24,  36,  20,  43,  27,  39,  23 },
> +    {  2,  50,  14,  62,   1,  49,  13,  61 },
> +    { 34,  18,  46,  30,  33,  17,  45,  29 },
> +    { 10,  58,   6,  54,   9,  57,   5,  53 },
> +    { 42,  26,  38,  22,  41,  25,  37,  21 },
> +};
> +
> +#define SQRT2 1.41421356237

M_SQRT2? Or does it affects bit-exactness?

[...]

No more comments from me without some serious study, but I believe it
should be fine assuming it is bitexact with the original filter.

"owdenoise" looks like a nice name to me, in case "ow" seems too
cryptic.
-- 
FFmpeg = Faithful & Fostering Martial Programmable Emblematic Gnome


More information about the ffmpeg-devel mailing list