• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1 /*
2  * Copyright (c) 2016 Clément Bœsch <u pkh me>
3  *
4  * This file is part of FFmpeg.
5  *
6  * FFmpeg is free software; you can redistribute it and/or
7  * modify it under the terms of the GNU Lesser General Public
8  * License as published by the Free Software Foundation; either
9  * version 2.1 of the License, or (at your option) any later version.
10  *
11  * FFmpeg is distributed in the hope that it will be useful,
12  * but WITHOUT ANY WARRANTY; without even the implied warranty of
13  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
14  * Lesser General Public License for more details.
15  *
16  * You should have received a copy of the GNU Lesser General Public
17  * License along with FFmpeg; if not, write to the Free Software
18  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
19  */
20 
21 /**
22  * @todo
23  * - better automatic defaults? see "Parameters" @ http://www.ipol.im/pub/art/2011/bcm_nlm/
24  * - temporal support (probably doesn't need any displacement according to
25  *   "Denoising image sequences does not require motion estimation")
26  * - Bayer pixel format support for at least raw photos? (DNG support would be
27  *   handy here)
28  * - FATE test (probably needs visual threshold test mechanism due to the use
29  *   of floats)
30  */
31 
32 #include "libavutil/avassert.h"
33 #include "libavutil/opt.h"
34 #include "libavutil/pixdesc.h"
35 #include "avfilter.h"
36 #include "formats.h"
37 #include "internal.h"
38 #include "vf_nlmeans.h"
39 #include "vf_nlmeans_init.h"
40 #include "video.h"
41 
42 typedef struct NLMeansContext {
43     const AVClass *class;
44     int nb_planes;
45     int chroma_w, chroma_h;
46     double pdiff_scale;                         // invert of the filtering parameter (sigma*10) squared
47     double sigma;                               // denoising strength
48     int patch_size,    patch_hsize;             // patch size and half size
49     int patch_size_uv, patch_hsize_uv;          // patch size and half size for chroma planes
50     int research_size,    research_hsize;       // research size and half size
51     int research_size_uv, research_hsize_uv;    // research size and half size for chroma planes
52     uint32_t *ii_orig;                          // integral image
53     uint32_t *ii;                               // integral image starting after the 0-line and 0-column
54     int ii_w, ii_h;                             // width and height of the integral image
55     ptrdiff_t ii_lz_32;                         // linesize in 32-bit units of the integral image
56     float *total_weight;                        // total weight for every pixel
57     float *sum;                                 // weighted sum for every pixel
58     int linesize;                               // sum and total_weight linesize
59     float *weight_lut;                          // lookup table mapping (scaled) patch differences to their associated weights
60     uint32_t max_meaningful_diff;               // maximum difference considered (if the patch difference is too high we ignore the pixel)
61     NLMeansDSPContext dsp;
62 } NLMeansContext;
63 
64 #define OFFSET(x) offsetof(NLMeansContext, x)
65 #define FLAGS AV_OPT_FLAG_FILTERING_PARAM|AV_OPT_FLAG_VIDEO_PARAM
66 static const AVOption nlmeans_options[] = {
67     { "s",  "denoising strength", OFFSET(sigma), AV_OPT_TYPE_DOUBLE, { .dbl = 1.0 }, 1.0, 30.0, FLAGS },
68     { "p",  "patch size",                   OFFSET(patch_size),    AV_OPT_TYPE_INT, { .i64 = 3*2+1 }, 0, 99, FLAGS },
69     { "pc", "patch size for chroma planes", OFFSET(patch_size_uv), AV_OPT_TYPE_INT, { .i64 = 0 },     0, 99, FLAGS },
70     { "r",  "research window",                   OFFSET(research_size),    AV_OPT_TYPE_INT, { .i64 = 7*2+1 }, 0, 99, FLAGS },
71     { "rc", "research window for chroma planes", OFFSET(research_size_uv), AV_OPT_TYPE_INT, { .i64 = 0 },     0, 99, FLAGS },
72     { NULL }
73 };
74 
75 AVFILTER_DEFINE_CLASS(nlmeans);
76 
77 static const enum AVPixelFormat pix_fmts[] = {
78     AV_PIX_FMT_YUV410P, AV_PIX_FMT_YUV411P,
79     AV_PIX_FMT_YUV420P, AV_PIX_FMT_YUV422P,
80     AV_PIX_FMT_YUV440P, AV_PIX_FMT_YUV444P,
81     AV_PIX_FMT_YUVJ444P, AV_PIX_FMT_YUVJ440P,
82     AV_PIX_FMT_YUVJ422P, AV_PIX_FMT_YUVJ420P,
83     AV_PIX_FMT_YUVJ411P,
84     AV_PIX_FMT_GRAY8, AV_PIX_FMT_GBRP,
85     AV_PIX_FMT_NONE
86 };
87 
88 /**
89  * Compute squared difference of an unsafe area (the zone nor s1 nor s2 could
90  * be readable).
91  *
92  * On the other hand, the line above dst and the column to its left are always
93  * readable.
94  *
95  * There is little point in having this function SIMDified as it is likely too
96  * complex and only handle small portions of the image.
97  *
98  * @param dst               integral image
99  * @param dst_linesize_32   integral image linesize (in 32-bit integers unit)
100  * @param startx            integral starting x position
101  * @param starty            integral starting y position
102  * @param src               source plane buffer
103  * @param linesize          source plane linesize
104  * @param offx              source offsetting in x
105  * @param offy              source offsetting in y
106  * @paran r                 absolute maximum source offsetting
107  * @param sw                source width
108  * @param sh                source height
109  * @param w                 width to compute
110  * @param h                 height to compute
111  */
compute_unsafe_ssd_integral_image(uint32_t * dst,ptrdiff_t dst_linesize_32,int startx,int starty,const uint8_t * src,ptrdiff_t linesize,int offx,int offy,int r,int sw,int sh,int w,int h)112 static inline void compute_unsafe_ssd_integral_image(uint32_t *dst, ptrdiff_t dst_linesize_32,
113                                                      int startx, int starty,
114                                                      const uint8_t *src, ptrdiff_t linesize,
115                                                      int offx, int offy, int r, int sw, int sh,
116                                                      int w, int h)
117 {
118     for (int y = starty; y < starty + h; y++) {
119         uint32_t acc = dst[y*dst_linesize_32 + startx - 1] - dst[(y-1)*dst_linesize_32 + startx - 1];
120         const int s1y = av_clip(y -  r,         0, sh - 1);
121         const int s2y = av_clip(y - (r + offy), 0, sh - 1);
122 
123         for (int x = startx; x < startx + w; x++) {
124             const int s1x = av_clip(x -  r,         0, sw - 1);
125             const int s2x = av_clip(x - (r + offx), 0, sw - 1);
126             const uint8_t v1 = src[s1y*linesize + s1x];
127             const uint8_t v2 = src[s2y*linesize + s2x];
128             const int d = v1 - v2;
129             acc += d * d;
130             dst[y*dst_linesize_32 + x] = dst[(y-1)*dst_linesize_32 + x] + acc;
131         }
132     }
133 }
134 
135 /*
136  * Compute the sum of squared difference integral image
137  * http://www.ipol.im/pub/art/2014/57/
138  * Integral Images for Block Matching - Gabriele Facciolo, Nicolas Limare, Enric Meinhardt-Llopis
139  *
140  * @param ii                integral image of dimension (w+e*2) x (h+e*2) with
141  *                          an additional zeroed top line and column already
142  *                          "applied" to the pointer value
143  * @param ii_linesize_32    integral image linesize (in 32-bit integers unit)
144  * @param src               source plane buffer
145  * @param linesize          source plane linesize
146  * @param offx              x-offsetting ranging in [-e;e]
147  * @param offy              y-offsetting ranging in [-e;e]
148  * @param w                 source width
149  * @param h                 source height
150  * @param e                 research padding edge
151  */
compute_ssd_integral_image(const NLMeansDSPContext * dsp,uint32_t * ii,ptrdiff_t ii_linesize_32,const uint8_t * src,ptrdiff_t linesize,int offx,int offy,int e,int w,int h)152 static void compute_ssd_integral_image(const NLMeansDSPContext *dsp,
153                                        uint32_t *ii, ptrdiff_t ii_linesize_32,
154                                        const uint8_t *src, ptrdiff_t linesize, int offx, int offy,
155                                        int e, int w, int h)
156 {
157     // ii has a surrounding padding of thickness "e"
158     const int ii_w = w + e*2;
159     const int ii_h = h + e*2;
160 
161     // we center the first source
162     const int s1x = e;
163     const int s1y = e;
164 
165     // 2nd source is the frame with offsetting
166     const int s2x = e + offx;
167     const int s2y = e + offy;
168 
169     // get the dimension of the overlapping rectangle where it is always safe
170     // to compare the 2 sources pixels
171     const int startx_safe = FFMAX(s1x, s2x);
172     const int starty_safe = FFMAX(s1y, s2y);
173     const int u_endx_safe = FFMIN(s1x + w, s2x + w); // unaligned
174     const int endy_safe   = FFMIN(s1y + h, s2y + h);
175 
176     // deduce the safe area width and height
177     const int safe_pw = (u_endx_safe - startx_safe) & ~0xf;
178     const int safe_ph = endy_safe - starty_safe;
179 
180     // adjusted end x position of the safe area after width of the safe area gets aligned
181     const int endx_safe = startx_safe + safe_pw;
182 
183     // top part where only one of s1 and s2 is still readable, or none at all
184     compute_unsafe_ssd_integral_image(ii, ii_linesize_32,
185                                       0, 0,
186                                       src, linesize,
187                                       offx, offy, e, w, h,
188                                       ii_w, starty_safe);
189 
190     // fill the left column integral required to compute the central
191     // overlapping one
192     compute_unsafe_ssd_integral_image(ii, ii_linesize_32,
193                                       0, starty_safe,
194                                       src, linesize,
195                                       offx, offy, e, w, h,
196                                       startx_safe, safe_ph);
197 
198     // main and safe part of the integral
199     av_assert1(startx_safe - s1x >= 0); av_assert1(startx_safe - s1x < w);
200     av_assert1(starty_safe - s1y >= 0); av_assert1(starty_safe - s1y < h);
201     av_assert1(startx_safe - s2x >= 0); av_assert1(startx_safe - s2x < w);
202     av_assert1(starty_safe - s2y >= 0); av_assert1(starty_safe - s2y < h);
203     if (safe_pw && safe_ph)
204         dsp->compute_safe_ssd_integral_image(ii + starty_safe*ii_linesize_32 + startx_safe, ii_linesize_32,
205                                              src + (starty_safe - s1y) * linesize + (startx_safe - s1x), linesize,
206                                              src + (starty_safe - s2y) * linesize + (startx_safe - s2x), linesize,
207                                              safe_pw, safe_ph);
208 
209     // right part of the integral
210     compute_unsafe_ssd_integral_image(ii, ii_linesize_32,
211                                       endx_safe, starty_safe,
212                                       src, linesize,
213                                       offx, offy, e, w, h,
214                                       ii_w - endx_safe, safe_ph);
215 
216     // bottom part where only one of s1 and s2 is still readable, or none at all
217     compute_unsafe_ssd_integral_image(ii, ii_linesize_32,
218                                       0, endy_safe,
219                                       src, linesize,
220                                       offx, offy, e, w, h,
221                                       ii_w, ii_h - endy_safe);
222 }
223 
config_input(AVFilterLink * inlink)224 static int config_input(AVFilterLink *inlink)
225 {
226     AVFilterContext *ctx = inlink->dst;
227     NLMeansContext *s = ctx->priv;
228     const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(inlink->format);
229     const int e = FFMAX(s->research_hsize, s->research_hsize_uv)
230                 + FFMAX(s->patch_hsize,    s->patch_hsize_uv);
231 
232     s->chroma_w = AV_CEIL_RSHIFT(inlink->w, desc->log2_chroma_w);
233     s->chroma_h = AV_CEIL_RSHIFT(inlink->h, desc->log2_chroma_h);
234     s->nb_planes = av_pix_fmt_count_planes(inlink->format);
235 
236     /* Allocate the integral image with extra edges of thickness "e"
237      *
238      *   +_+-------------------------------+
239      *   |0|0000000000000000000000000000000|
240      *   +-x-------------------------------+
241      *   |0|\    ^                         |
242      *   |0| ii  | e                       |
243      *   |0|     v                         |
244      *   |0|   +-----------------------+   |
245      *   |0|   |                       |   |
246      *   |0|<->|                       |   |
247      *   |0| e |                       |   |
248      *   |0|   |                       |   |
249      *   |0|   +-----------------------+   |
250      *   |0|                               |
251      *   |0|                               |
252      *   |0|                               |
253      *   +-+-------------------------------+
254      */
255     s->ii_w = inlink->w + e*2;
256     s->ii_h = inlink->h + e*2;
257 
258     // align to 4 the linesize, "+1" is for the space of the left 0-column
259     s->ii_lz_32 = FFALIGN(s->ii_w + 1, 4);
260 
261     // "+1" is for the space of the top 0-line
262     s->ii_orig = av_calloc(s->ii_h + 1, s->ii_lz_32 * sizeof(*s->ii_orig));
263     if (!s->ii_orig)
264         return AVERROR(ENOMEM);
265 
266     // skip top 0-line and left 0-column
267     s->ii = s->ii_orig + s->ii_lz_32 + 1;
268 
269     // allocate weighted average for every pixel
270     s->linesize = inlink->w + 100;
271     s->total_weight = av_malloc_array(s->linesize, inlink->h * sizeof(*s->total_weight));
272     s->sum = av_malloc_array(s->linesize, inlink->h * sizeof(*s->sum));
273     if (!s->total_weight || !s->sum)
274         return AVERROR(ENOMEM);
275 
276     return 0;
277 }
278 
279 struct thread_data {
280     const uint8_t *src;
281     ptrdiff_t src_linesize;
282     int startx, starty;
283     int endx, endy;
284     const uint32_t *ii_start;
285     int p;
286 };
287 
nlmeans_slice(AVFilterContext * ctx,void * arg,int jobnr,int nb_jobs)288 static int nlmeans_slice(AVFilterContext *ctx, void *arg, int jobnr, int nb_jobs)
289 {
290     NLMeansContext *s = ctx->priv;
291     const uint32_t max_meaningful_diff = s->max_meaningful_diff;
292     const struct thread_data *td = arg;
293     const ptrdiff_t src_linesize = td->src_linesize;
294     const int process_h = td->endy - td->starty;
295     const int slice_start = (process_h *  jobnr   ) / nb_jobs;
296     const int slice_end   = (process_h * (jobnr+1)) / nb_jobs;
297     const int starty = td->starty + slice_start;
298     const int endy   = td->starty + slice_end;
299     const int p = td->p;
300     const uint32_t *ii = td->ii_start + (starty - p - 1) * s->ii_lz_32 - p - 1;
301     const int dist_b = 2*p + 1;
302     const int dist_d = dist_b * s->ii_lz_32;
303     const int dist_e = dist_d + dist_b;
304     const float *const weight_lut = s->weight_lut;
305     NLMeansDSPContext *dsp = &s->dsp;
306 
307     for (int y = starty; y < endy; y++) {
308         const uint8_t *const src = td->src + y*src_linesize;
309         float *total_weight = s->total_weight + y*s->linesize;
310         float *sum = s->sum + y*s->linesize;
311         const uint32_t *const iia = ii;
312         const uint32_t *const iib = ii + dist_b;
313         const uint32_t *const iid = ii + dist_d;
314         const uint32_t *const iie = ii + dist_e;
315 
316         dsp->compute_weights_line(iia, iib, iid, iie, src, total_weight, sum,
317                                   weight_lut, max_meaningful_diff,
318                                   td->startx, td->endx);
319         ii += s->ii_lz_32;
320     }
321     return 0;
322 }
323 
weight_averages(uint8_t * dst,ptrdiff_t dst_linesize,const uint8_t * src,ptrdiff_t src_linesize,float * total_weight,float * sum,ptrdiff_t linesize,int w,int h)324 static void weight_averages(uint8_t *dst, ptrdiff_t dst_linesize,
325                             const uint8_t *src, ptrdiff_t src_linesize,
326                             float *total_weight, float *sum, ptrdiff_t linesize,
327                             int w, int h)
328 {
329     for (int y = 0; y < h; y++) {
330         for (int x = 0; x < w; x++) {
331             // Also weight the centered pixel
332             total_weight[x] += 1.f;
333             sum[x] += 1.f * src[x];
334             dst[x] = av_clip_uint8(sum[x] / total_weight[x] + 0.5f);
335         }
336         dst += dst_linesize;
337         src += src_linesize;
338         total_weight += linesize;
339         sum += linesize;
340     }
341 }
342 
nlmeans_plane(AVFilterContext * ctx,int w,int h,int p,int r,uint8_t * dst,ptrdiff_t dst_linesize,const uint8_t * src,ptrdiff_t src_linesize)343 static int nlmeans_plane(AVFilterContext *ctx, int w, int h, int p, int r,
344                          uint8_t *dst, ptrdiff_t dst_linesize,
345                          const uint8_t *src, ptrdiff_t src_linesize)
346 {
347     NLMeansContext *s = ctx->priv;
348     /* patches center points cover the whole research window so the patches
349      * themselves overflow the research window */
350     const int e = r + p;
351     /* focus an integral pointer on the centered image (s1) */
352     const uint32_t *centered_ii = s->ii + e*s->ii_lz_32 + e;
353 
354     memset(s->total_weight, 0, s->linesize * h * sizeof(*s->total_weight));
355     memset(s->sum, 0, s->linesize * h * sizeof(*s->sum));
356 
357     for (int offy = -r; offy <= r; offy++) {
358         for (int offx = -r; offx <= r; offx++) {
359             if (offx || offy) {
360                 struct thread_data td = {
361                     .src          = src + offy*src_linesize + offx,
362                     .src_linesize = src_linesize,
363                     .startx       = FFMAX(0, -offx),
364                     .starty       = FFMAX(0, -offy),
365                     .endx         = FFMIN(w, w - offx),
366                     .endy         = FFMIN(h, h - offy),
367                     .ii_start     = centered_ii + offy*s->ii_lz_32 + offx,
368                     .p            = p,
369                 };
370 
371                 compute_ssd_integral_image(&s->dsp, s->ii, s->ii_lz_32,
372                                            src, src_linesize,
373                                            offx, offy, e, w, h);
374                 ff_filter_execute(ctx, nlmeans_slice, &td, NULL,
375                                   FFMIN(td.endy - td.starty, ff_filter_get_nb_threads(ctx)));
376             }
377         }
378     }
379 
380     weight_averages(dst, dst_linesize, src, src_linesize,
381                     s->total_weight, s->sum, s->linesize, w, h);
382 
383     return 0;
384 }
385 
filter_frame(AVFilterLink * inlink,AVFrame * in)386 static int filter_frame(AVFilterLink *inlink, AVFrame *in)
387 {
388     AVFilterContext *ctx = inlink->dst;
389     NLMeansContext *s = ctx->priv;
390     AVFilterLink *outlink = ctx->outputs[0];
391 
392     AVFrame *out = ff_get_video_buffer(outlink, outlink->w, outlink->h);
393     if (!out) {
394         av_frame_free(&in);
395         return AVERROR(ENOMEM);
396     }
397     av_frame_copy_props(out, in);
398 
399     for (int i = 0; i < s->nb_planes; i++) {
400         const int w = i ? s->chroma_w          : inlink->w;
401         const int h = i ? s->chroma_h          : inlink->h;
402         const int p = i ? s->patch_hsize_uv    : s->patch_hsize;
403         const int r = i ? s->research_hsize_uv : s->research_hsize;
404         nlmeans_plane(ctx, w, h, p, r,
405                       out->data[i], out->linesize[i],
406                       in->data[i],  in->linesize[i]);
407     }
408 
409     av_frame_free(&in);
410     return ff_filter_frame(outlink, out);
411 }
412 
413 #define CHECK_ODD_FIELD(field, name) do {                       \
414     if (!(s->field & 1)) {                                      \
415         s->field |= 1;                                          \
416         av_log(ctx, AV_LOG_WARNING, name " size must be odd, "  \
417                "setting it to %d\n", s->field);                 \
418     }                                                           \
419 } while (0)
420 
init(AVFilterContext * ctx)421 static av_cold int init(AVFilterContext *ctx)
422 {
423     NLMeansContext *s = ctx->priv;
424     const double h = s->sigma * 10.;
425 
426     s->pdiff_scale = 1. / (h * h);
427     s->max_meaningful_diff = log(255.) / s->pdiff_scale;
428     s->weight_lut = av_calloc(s->max_meaningful_diff + 1, sizeof(*s->weight_lut));
429     if (!s->weight_lut)
430         return AVERROR(ENOMEM);
431     for (int i = 0; i < s->max_meaningful_diff; i++)
432         s->weight_lut[i] = exp(-i * s->pdiff_scale);
433 
434     CHECK_ODD_FIELD(research_size,   "Luma research window");
435     CHECK_ODD_FIELD(patch_size,      "Luma patch");
436 
437     if (!s->research_size_uv) s->research_size_uv = s->research_size;
438     if (!s->patch_size_uv)    s->patch_size_uv    = s->patch_size;
439 
440     CHECK_ODD_FIELD(research_size_uv, "Chroma research window");
441     CHECK_ODD_FIELD(patch_size_uv,    "Chroma patch");
442 
443     s->research_hsize    = s->research_size    / 2;
444     s->research_hsize_uv = s->research_size_uv / 2;
445     s->patch_hsize       = s->patch_size       / 2;
446     s->patch_hsize_uv    = s->patch_size_uv    / 2;
447 
448     av_log(ctx, AV_LOG_DEBUG, "Research window: %dx%d / %dx%d, patch size: %dx%d / %dx%d\n",
449            s->research_size, s->research_size, s->research_size_uv, s->research_size_uv,
450            s->patch_size,    s->patch_size,    s->patch_size_uv,    s->patch_size_uv);
451 
452     ff_nlmeans_init(&s->dsp);
453 
454     return 0;
455 }
456 
uninit(AVFilterContext * ctx)457 static av_cold void uninit(AVFilterContext *ctx)
458 {
459     NLMeansContext *s = ctx->priv;
460     av_freep(&s->weight_lut);
461     av_freep(&s->ii_orig);
462     av_freep(&s->total_weight);
463     av_freep(&s->sum);
464 }
465 
466 static const AVFilterPad nlmeans_inputs[] = {
467     {
468         .name         = "default",
469         .type         = AVMEDIA_TYPE_VIDEO,
470         .config_props = config_input,
471         .filter_frame = filter_frame,
472     },
473 };
474 
475 static const AVFilterPad nlmeans_outputs[] = {
476     {
477         .name = "default",
478         .type = AVMEDIA_TYPE_VIDEO,
479     },
480 };
481 
482 const AVFilter ff_vf_nlmeans = {
483     .name          = "nlmeans",
484     .description   = NULL_IF_CONFIG_SMALL("Non-local means denoiser."),
485     .priv_size     = sizeof(NLMeansContext),
486     .init          = init,
487     .uninit        = uninit,
488     FILTER_INPUTS(nlmeans_inputs),
489     FILTER_OUTPUTS(nlmeans_outputs),
490     FILTER_PIXFMTS_ARRAY(pix_fmts),
491     .priv_class    = &nlmeans_class,
492     .flags         = AVFILTER_FLAG_SUPPORT_TIMELINE_GENERIC | AVFILTER_FLAG_SLICE_THREADS,
493 };
494