[FFmpeg-devel] [PATCH] lavfi/RFC: add dctdnoiz filter
Clément Bœsch
ubitux at gmail.com
Sat Apr 27 17:19:21 CEST 2013
Simple and extremely slow DCT denoiser.
TODO: doc, minor bump, Changelog
---
Basically a brain dead implementation of
www.ipol.im/pub/art/2011/ys-dct/gg
In addition there is an overlap parameter (which needs some adjustment
to deal properly with borders), and an optional expression allowing more
advanced coefficient modifications than a hard threshold like proposed
in the paper.
Equivalence between sigma and expression parameters:
dctdnoiz=20 <=> dctdnoiz=e='gte(c,20*3)'
With the examples from the paper:
noise 15: http://i.imgur.com/UBDaAlX.jpg
noise 30: http://i.imgur.com/yEfwC21.jpg
---
configure | 1 +
libavfilter/Makefile | 1 +
libavfilter/allfilters.c | 1 +
libavfilter/vf_dctdnoiz.c | 389 ++++++++++++++++++++++++++++++++++++++++++++++
4 files changed, 392 insertions(+)
create mode 100644 libavfilter/vf_dctdnoiz.c
diff --git a/configure b/configure
index 648db2f..5158646 100755
--- a/configure
+++ b/configure
@@ -2120,6 +2120,7 @@ blackframe_filter_deps="gpl"
boxblur_filter_deps="gpl"
colormatrix_filter_deps="gpl"
cropdetect_filter_deps="gpl"
+dctdnoiz_filter_deps="avcodec"
delogo_filter_deps="gpl"
deshake_filter_deps="avcodec"
deshake_filter_select="dsputil"
diff --git a/libavfilter/Makefile b/libavfilter/Makefile
index 358a4b6..43b3eae 100644
--- a/libavfilter/Makefile
+++ b/libavfilter/Makefile
@@ -114,6 +114,7 @@ OBJS-$(CONFIG_COPY_FILTER) += vf_copy.o
OBJS-$(CONFIG_CROP_FILTER) += vf_crop.o
OBJS-$(CONFIG_CROPDETECT_FILTER) += vf_cropdetect.o
OBJS-$(CONFIG_CURVES_FILTER) += vf_curves.o
+OBJS-$(CONFIG_DCTDNOIZ_FILTER) += vf_dctdnoiz.o
OBJS-$(CONFIG_DECIMATE_FILTER) += vf_decimate.o
OBJS-$(CONFIG_DELOGO_FILTER) += vf_delogo.o
OBJS-$(CONFIG_DESHAKE_FILTER) += vf_deshake.o
diff --git a/libavfilter/allfilters.c b/libavfilter/allfilters.c
index 97e558b..8f5dc63 100644
--- a/libavfilter/allfilters.c
+++ b/libavfilter/allfilters.c
@@ -112,6 +112,7 @@ void avfilter_register_all(void)
REGISTER_FILTER(CROP, crop, vf);
REGISTER_FILTER(CROPDETECT, cropdetect, vf);
REGISTER_FILTER(CURVES, curves, vf);
+ REGISTER_FILTER(DCTDNOIZ, dctdnoiz, vf);
REGISTER_FILTER(DECIMATE, decimate, vf);
REGISTER_FILTER(DELOGO, delogo, vf);
REGISTER_FILTER(DESHAKE, deshake, vf);
diff --git a/libavfilter/vf_dctdnoiz.c b/libavfilter/vf_dctdnoiz.c
new file mode 100644
index 0000000..5841e7c
--- /dev/null
+++ b/libavfilter/vf_dctdnoiz.c
@@ -0,0 +1,389 @@
+/*
+ * 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
+ */
+
+/**
+ * A simple, relatively efficient and extremely slow DCT image denoiser.
+ * @see http://www.ipol.im/pub/art/2011/ys-dct/
+ */
+
+#include "libavcodec/avfft.h"
+#include "libavutil/eval.h"
+#include "libavutil/opt.h"
+#include "drawutils.h"
+#include "internal.h"
+
+#define NBITS 4
+#define BSIZE (1<<(NBITS))
+
+static const char *const var_names[] = { "c", NULL };
+enum { VAR_C, VAR_VARS_NB };
+
+typedef struct {
+ const AVClass *class;
+
+ /* coefficient factor expression */
+ char *expr_str;
+ AVExpr *expr;
+ double var_values[VAR_VARS_NB];
+
+ float sigma; // used when no expression are st
+ float th; // threshold (3*sigma)
+ float color_dct[3][3]; // 3x3 DCT for color decorrelation
+ float *cbuf[2][3]; // two planar rgb color buffers
+ float *weights; // dct coeff are cumulated with overlapping; these values are used for averaging
+ int p_linesize; // line sizes for color and weights
+ int overlap; // number of block overlapping pixels
+ int step; // block step increment (BSIZE - overlap)
+ DCTContext *dct, *idct; // DCT and inverse DCT contexts
+ float *block, *tmp_block; // two BSIZE x BSIZE block buffers
+} DCTdnoizContext;
+
+#define OFFSET(x) offsetof(DCTdnoizContext, x)
+#define FLAGS AV_OPT_FLAG_FILTERING_PARAM|AV_OPT_FLAG_VIDEO_PARAM
+static const AVOption dctdnoiz_options[] = {
+ { "s", "set noise sigma constant", OFFSET(sigma), AV_OPT_TYPE_FLOAT, {.dbl=0}, 0, 999, .flags = FLAGS },
+ { "overlap", "set number of block overlapping pixels", OFFSET(overlap), AV_OPT_TYPE_INT, {.i64=(1<<NBITS)-1}, 0, (1<<NBITS)-1, .flags = FLAGS },
+ { "e", "set coefficient factor expression", OFFSET(expr_str), AV_OPT_TYPE_STRING, {.str=NULL}, .flags = FLAGS },
+ { NULL }
+};
+
+AVFILTER_DEFINE_CLASS(dctdnoiz);
+
+static float *dct_block(DCTdnoizContext *ctx, const float *src, int src_linesize)
+{
+ int x, y;
+ float *column;
+
+ for (y = 0; y < BSIZE; y++) {
+ float *line = ctx->block;
+
+ memcpy(line, src, BSIZE * sizeof(*line));
+ src += src_linesize;
+ av_dct_calc(ctx->dct, line);
+
+ column = ctx->tmp_block + y;
+ for (x = 0; x < BSIZE; x++) {
+ *line *= x == 0 ? 1. / sqrt(BSIZE) : sqrt(2. / BSIZE);
+ *column = *line++;
+ column += BSIZE;
+ }
+ }
+
+ column = ctx->tmp_block;
+ for (x = 0; x < BSIZE; x++) {
+ av_dct_calc(ctx->dct, column);
+ for (y = 0; y < BSIZE; y++)
+ column[y] *= y == 0 ? 1. / sqrt(BSIZE) : sqrt(2. / BSIZE);
+ column += BSIZE;
+ }
+
+ for (y = 0; y < BSIZE; y++)
+ for (x = 0; x < BSIZE; x++)
+ ctx->block[y*BSIZE + x] = ctx->tmp_block[x*BSIZE + y];
+
+ return ctx->block;
+}
+
+static void idct_block(DCTdnoizContext *ctx, float *dst, int dst_linesize)
+{
+ int x, y;
+ float *block = ctx->block;
+ float *tmp = ctx->tmp_block;
+
+ for (y = 0; y < BSIZE; y++) {
+ for (x = 0; x < BSIZE; x++)
+ block[x] *= x == 0 ? sqrt(BSIZE) : 1./sqrt(2. / BSIZE);
+ av_dct_calc(ctx->idct, block);
+ block += BSIZE;
+ }
+
+ block = ctx->block;
+ for (y = 0; y < BSIZE; y++) {
+ for (x = 0; x < BSIZE; x++) {
+ tmp[x] = block[x*BSIZE + y];
+ tmp[x] *= x == 0 ? sqrt(BSIZE) : 1./sqrt(2. / BSIZE);
+ }
+ av_dct_calc(ctx->idct, tmp);
+ for (x = 0; x < BSIZE; x++)
+ dst[x*dst_linesize + y] += tmp[x];
+ }
+}
+
+static int config_input(AVFilterLink *inlink)
+{
+ DCTdnoizContext *ddn = inlink->dst->priv;
+ const int linesize = FFALIGN(inlink->w, 16);
+ int i, x, y, bx, by, *iweights;
+ const float dct_3x3[3][3] = {
+ { 1./sqrt(3), 1./sqrt(3), 1./sqrt(3) },
+ { 1./sqrt(2), 0, -1./sqrt(2) },
+ { 1./sqrt(6), -2./sqrt(6), 1./sqrt(6) },
+ };
+ uint8_t rgba_map[4];
+
+ ff_fill_rgba_map(rgba_map, inlink->format);
+ for (y = 0; y < 3; y++)
+ for (x = 0; x < 3; x++)
+ ddn->color_dct[y][x] = dct_3x3[rgba_map[y]][rgba_map[x]];
+
+ ddn->p_linesize = linesize;
+ for (i = 0; i < 2; i++) {
+ ddn->cbuf[i][0] = av_malloc(linesize * inlink->h * sizeof(*ddn->cbuf[i][0]));
+ ddn->cbuf[i][1] = av_malloc(linesize * inlink->h * sizeof(*ddn->cbuf[i][1]));
+ ddn->cbuf[i][2] = av_malloc(linesize * inlink->h * sizeof(*ddn->cbuf[i][2]));
+ if (!ddn->cbuf[i][0] || !ddn->cbuf[i][1] || !ddn->cbuf[i][2])
+ return AVERROR(ENOMEM);
+ }
+
+ ddn->weights = av_malloc(inlink->h * linesize * sizeof(*ddn->weights));
+ if (!ddn->weights)
+ return AVERROR(ENOMEM);
+ iweights = av_calloc(inlink->h, linesize * sizeof(*iweights));
+ if (!iweights)
+ return AVERROR(ENOMEM);
+ for (y = 0; y < inlink->h - BSIZE + 1; y += ddn->step)
+ for (x = 0; x < inlink->w - BSIZE + 1; x += ddn->step)
+ for (by = 0; by < BSIZE; by++)
+ for (bx = 0; bx < BSIZE; bx++)
+ iweights[(y + by)*linesize + x + bx]++;
+ for (y = 0; y < inlink->h; y++)
+ for (x = 0; x < inlink->w; x++)
+ ddn->weights[y*linesize + x] = 1. / iweights[y*linesize + x];
+ av_free(iweights);
+
+ return 0;
+}
+
+static av_cold int init(AVFilterContext *ctx)
+{
+ DCTdnoizContext *ddn = ctx->priv;
+
+ if (ddn->expr_str) {
+ int ret = av_expr_parse(&ddn->expr, ddn->expr_str, var_names,
+ NULL, NULL, NULL, NULL, 0, ctx);
+ if (ret < 0)
+ return ret;
+ }
+
+ ddn->th = ddn->sigma * 3.;
+ ddn->step = BSIZE - ddn->overlap;
+ ddn->dct = av_dct_init(NBITS, DCT_II);
+ ddn->idct = av_dct_init(NBITS, DCT_III);
+ ddn->block = av_malloc(BSIZE * BSIZE * sizeof(*ddn->block));
+ ddn->tmp_block = av_malloc(BSIZE * BSIZE * sizeof(*ddn->tmp_block));
+
+ if (!ddn->dct || !ddn->idct || !ddn->tmp_block || !ddn->block)
+ return AVERROR(ENOMEM);
+
+ return 0;
+}
+
+static int query_formats(AVFilterContext *ctx)
+{
+ static const enum AVPixelFormat pix_fmts[] = {
+ AV_PIX_FMT_BGR24, AV_PIX_FMT_RGB24,
+ AV_PIX_FMT_NONE
+ };
+ ff_set_common_formats(ctx, ff_make_format_list(pix_fmts));
+ return 0;
+}
+
+static void color_decorrelation(float dct3ch[3][3], float **dst, int dst_linesize,
+ const uint8_t *src, int src_linesize, int w, int h)
+{
+ int x, y;
+ float *dstp_r = dst[0];
+ float *dstp_g = dst[1];
+ float *dstp_b = dst[2];
+
+ for (y = 0; y < h; y++) {
+ const uint8_t *srcp = src;
+
+ for (x = 0; x < w; x++) {
+ dstp_r[x] = srcp[0] * dct3ch[0][0] + srcp[1] * dct3ch[0][1] + srcp[2] * dct3ch[0][2];
+ dstp_g[x] = srcp[0] * dct3ch[1][0] + srcp[1] * dct3ch[1][1] + srcp[2] * dct3ch[1][2];
+ dstp_b[x] = srcp[0] * dct3ch[2][0] + srcp[1] * dct3ch[2][1] + srcp[2] * dct3ch[2][2];
+ srcp += 3;
+ }
+ src += src_linesize;
+ dstp_r += dst_linesize;
+ dstp_g += dst_linesize;
+ dstp_b += dst_linesize;
+ }
+}
+
+static void color_correlation(float dct3ch[3][3], uint8_t *dst, int dst_linesize,
+ float **src, int src_linesize, int w, int h)
+{
+ int x, y;
+ const float *src_r = src[0];
+ const float *src_g = src[1];
+ const float *src_b = src[2];
+
+ for (y = 0; y < h; y++) {
+ uint8_t *dstp = dst;
+
+ for (x = 0; x < w; x++) {
+ dstp[0] = av_clip_uint8(src_r[x] * dct3ch[0][0] + src_g[x] * dct3ch[1][0] + src_b[x] * dct3ch[2][0]);
+ dstp[1] = av_clip_uint8(src_r[x] * dct3ch[0][1] + src_g[x] * dct3ch[1][1] + src_b[x] * dct3ch[2][1]);
+ dstp[2] = av_clip_uint8(src_r[x] * dct3ch[0][2] + src_g[x] * dct3ch[1][2] + src_b[x] * dct3ch[2][2]);
+ dstp += 3;
+ }
+ dst += dst_linesize;
+ src_r += src_linesize;
+ src_g += src_linesize;
+ src_b += src_linesize;
+ }
+}
+
+static void filter_plane(AVFilterContext *ctx,
+ float *dst, int dst_linesize,
+ const float *src, int src_linesize,
+ int w, int h)
+{
+ int x, y, bx, by;
+ DCTdnoizContext *ddn = ctx->priv;
+ float *dst0 = dst;
+ const float *weights = ddn->weights;
+
+ // reset block sums
+ memset(dst, 0, h * dst_linesize * sizeof(*dst));
+
+ // block dct sums
+ for (y = 0; y < h - BSIZE + 1; y += ddn->step) {
+ for (x = 0; x < w - BSIZE + 1; x += ddn->step) {
+ float *ftb = dct_block(ddn, src + x, src_linesize);
+
+ if (ddn->expr) {
+ for (by = 0; by < BSIZE; by++) {
+ for (bx = 0; bx < BSIZE; bx++) {
+ ddn->var_values[VAR_C] = FFABS(*ftb);
+ *ftb++ *= av_expr_eval(ddn->expr, ddn->var_values, ddn);
+ }
+ }
+ } else {
+ for (by = 0; by < BSIZE; by++) {
+ for (bx = 0; bx < BSIZE; bx++) {
+ if (FFABS(*ftb) < ddn->th)
+ *ftb = 0;
+ ftb++;
+ }
+ }
+ }
+ idct_block(ddn, dst + x, dst_linesize);
+ }
+ src += ddn->step * src_linesize;
+ dst += ddn->step * dst_linesize;
+ }
+
+ // average blocks
+ dst = dst0;
+ for (y = 0; y < h; y++) {
+ for (x = 0; x < w; x++)
+ dst[x] *= weights[x];
+ dst += dst_linesize;
+ weights += dst_linesize;
+ }
+}
+
+static int filter_frame(AVFilterLink *inlink, AVFrame *in)
+{
+ AVFilterContext *ctx = inlink->dst;
+ DCTdnoizContext *ddn = ctx->priv;
+ AVFilterLink *outlink = inlink->dst->outputs[0];
+ int direct, plane;
+ AVFrame *out;
+
+ if (av_frame_is_writable(in)) {
+ direct = 1;
+ out = in;
+ } else {
+ direct = 0;
+ out = ff_get_video_buffer(outlink, outlink->w, outlink->h);
+ if (!out) {
+ av_frame_free(&in);
+ return AVERROR(ENOMEM);
+ }
+ av_frame_copy_props(out, in);
+ }
+
+ color_decorrelation(ddn->color_dct, ddn->cbuf[0], ddn->p_linesize,
+ in->data[0], in->linesize[0], inlink->w, inlink->h);
+ for (plane = 0; plane < 3; plane++)
+ filter_plane(ctx, ddn->cbuf[1][plane], ddn->p_linesize,
+ ddn->cbuf[0][plane], ddn->p_linesize,
+ inlink->w, inlink->h);
+ color_correlation(ddn->color_dct, out->data[0], out->linesize[0],
+ ddn->cbuf[1], ddn->p_linesize, inlink->w, inlink->h);
+
+ if (!direct)
+ av_frame_free(&in);
+
+ return ff_filter_frame(outlink, out);
+}
+
+static av_cold void uninit(AVFilterContext *ctx)
+{
+ int i;
+ DCTdnoizContext *ddn = ctx->priv;
+
+ av_dct_end(ddn->dct);
+ av_dct_end(ddn->idct);
+ av_free(ddn->block);
+ av_free(ddn->tmp_block);
+ av_free(ddn->weights);
+ for (i = 0; i < 2; i++) {
+ av_free(ddn->cbuf[i][0]);
+ av_free(ddn->cbuf[i][1]);
+ av_free(ddn->cbuf[i][2]);
+ }
+ av_expr_free(ddn->expr);
+}
+
+static const AVFilterPad dctdnoiz_inputs[] = {
+ {
+ .name = "default",
+ .type = AVMEDIA_TYPE_VIDEO,
+ .filter_frame = filter_frame,
+ .config_props = config_input,
+ },
+ { NULL }
+};
+
+static const AVFilterPad dctdnoiz_outputs[] = {
+ {
+ .name = "default",
+ .type = AVMEDIA_TYPE_VIDEO,
+ },
+ { NULL }
+};
+
+AVFilter avfilter_vf_dctdnoiz = {
+ .name = "dctdnoiz",
+ .description = NULL_IF_CONFIG_SMALL("Denoise frames using 2D DCT."),
+ .priv_size = sizeof(DCTdnoizContext),
+ .init = init,
+ .uninit = uninit,
+ .query_formats = query_formats,
+ .inputs = dctdnoiz_inputs,
+ .outputs = dctdnoiz_outputs,
+ .priv_class = &dctdnoiz_class,
+ .flags = AVFILTER_FLAG_SUPPORT_TIMELINE,
+};
--
1.8.2.1
More information about the ffmpeg-devel
mailing list