ffmpeg/libavfilter/vf_psnr.c

   1 /*
   2  * Copyright (c) 2011 Roger Pau Monné <roger.pau@entel.upc.edu>
   3  * Copyright (c) 2011 Stefano Sabatini
   4  * Copyright (c) 2013 Paul B Mahol
   5  *
   6  * This file is part of FFmpeg.
   7  *
   8  * FFmpeg is free software; you can redistribute it and/or
   9  * modify it under the terms of the GNU Lesser General Public
  10  * License as published by the Free Software Foundation; either
  11  * version 2.1 of the License, or (at your option) any later version.
  12  *
  13  * FFmpeg is distributed in the hope that it will be useful,
  14  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  15  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  16  * Lesser General Public License for more details.
  17  *
  18  * You should have received a copy of the GNU Lesser General Public
  19  * License along with FFmpeg; if not, write to the Free Software
  20  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  21  */
  22
  23 /**
  24  * @file
  25  * Caculate the PSNR between two input videos.
  26  */
  27
  28 #include "libavutil/opt.h"
  29 #include "libavutil/pixdesc.h"
  30 #include "avfilter.h"
  31 #include "dualinput.h"
  32 #include "drawutils.h"
  33 #include "formats.h"
  34 #include "internal.h"
  35 #include "video.h"
  36
  37 typedef struct PSNRContext {
  38     const AVClass *class;
  39     FFDualInputContext dinput;
  40     double mse, min_mse, max_mse;
  41     uint64_t nb_frames;
  42     FILE *stats_file;
  43     char *stats_file_str;
  44     int max[4], average_max;
  45     int is_rgb;
  46     uint8_t rgba_map[4];
  47     char comps[4];
  48     int nb_components;
  49     int planewidth[4];
  50     int planeheight[4];
  51
  52     void (*compute_mse)(struct PSNRContext *s,
  53                         const uint8_t *m[4], const int ml[4],
  54                         const uint8_t *r[4], const int rl[4],
  55                         int w, int h, double mse[4]);
  56 } PSNRContext;
  57
  58 #define OFFSET(x) offsetof(PSNRContext, x)
  59 #define FLAGS AV_OPT_FLAG_FILTERING_PARAM|AV_OPT_FLAG_VIDEO_PARAM
  60
  61 static const AVOption psnr_options[] = {
  62     {"stats_file", "Set file where to store per-frame difference information", OFFSET(stats_file_str), AV_OPT_TYPE_STRING, {.str=NULL}, 0, 0, FLAGS },
  63     {"f",          "Set file where to store per-frame difference information", OFFSET(stats_file_str), AV_OPT_TYPE_STRING, {.str=NULL}, 0, 0, FLAGS },
  64     { NULL }
  65 };
  66
  67 AVFILTER_DEFINE_CLASS(psnr);
  68
  69 static inline unsigned pow2(unsigned base)
  70 {
  71     return base*base;
  72 }
  73
  74 static inline double get_psnr(double mse, uint64_t nb_frames, int max)
  75 {
  76     return 10.0 * log(pow2(max) / (mse / nb_frames)) / log(10.0);
  77 }
  78
  79 static inline
  80 void compute_images_mse(PSNRContext *s,
  81                         const uint8_t *main_data[4], const int main_linesizes[4],
  82                         const uint8_t *ref_data[4], const int ref_linesizes[4],
  83                         int w, int h, double mse[4])
  84 {
  85     int i, c, j;
  86
  87     for (c = 0; c < s->nb_components; c++) {
  88         const int outw = s->planewidth[c];
  89         const int outh = s->planeheight[c];
  90         const uint8_t *main_line = main_data[c];
  91         const uint8_t *ref_line = ref_data[c];
  92         const int ref_linesize = ref_linesizes[c];
  93         const int main_linesize = main_linesizes[c];
  94         uint64_t m = 0;
  95
  96         for (i = 0; i < outh; i++) {
  97             int m2 = 0;
  98             for (j = 0; j < outw; j++)
  99                 m2 += pow2(main_line[j] - ref_line[j]);
 100             m += m2;
 101             ref_line += ref_linesize;
 102             main_line += main_linesize;
 103         }
 104         mse[c] = m / (double)(outw * outh);
 105     }
 106 }
 107
 108 static inline
 109 void compute_images_mse_16bit(PSNRContext *s,
 110                         const uint8_t *main_data[4], const int main_linesizes[4],
 111                         const uint8_t *ref_data[4], const int ref_linesizes[4],
 112                         int w, int h, double mse[4])
 113 {
 114     int i, c, j;
 115
 116     for (c = 0; c < s->nb_components; c++) {
 117         const int outw = s->planewidth[c];
 118         const int outh = s->planeheight[c];
 119         const uint16_t *main_line = (uint16_t *)main_data[c];
 120         const uint16_t *ref_line = (uint16_t *)ref_data[c];
 121         const int ref_linesize = ref_linesizes[c] / 2;
 122         const int main_linesize = main_linesizes[c] / 2;
 123         uint64_t m = 0;
 124
 125         for (i = 0; i < outh; i++) {
 126             for (j = 0; j < outw; j++)
 127                 m += pow2(main_line[j] - ref_line[j]);
 128             ref_line += ref_linesize;
 129             main_line += main_linesize;
 130         }
 131         mse[c] = m / (double)(outw * outh);
 132     }
 133 }
 134
 135 static void set_meta(AVDictionary **metadata, const char *key, char comp, float d)
 136 {
 137     char value[128];
 138     snprintf(value, sizeof(value), "%0.2f", d);
 139     if (comp) {
 140         char key2[128];
 141         snprintf(key2, sizeof(key2), "%s%c", key, comp);
 142         av_dict_set(metadata, key2, value, 0);
 143     } else {
 144         av_dict_set(metadata, key, value, 0);
 145     }
 146 }
 147
 148 static AVFrame *do_psnr(AVFilterContext *ctx, AVFrame *main,
 149                         const AVFrame *ref)
 150 {
 151     PSNRContext *s = ctx->priv;
 152     double comp_mse[4], mse = 0;
 153     int j, c;
 154     AVDictionary **metadata = avpriv_frame_get_metadatap(main);
 155
 156     s->compute_mse(s, (const uint8_t **)main->data, main->linesize,
 157                       (const uint8_t **)ref->data, ref->linesize,
 158                        main->width, main->height, comp_mse);
 159
 160     for (j = 0; j < s->nb_components; j++)
 161         mse += comp_mse[j];
 162     mse /= s->nb_components;
 163
 164     s->min_mse = FFMIN(s->min_mse, mse);
 165     s->max_mse = FFMAX(s->max_mse, mse);
 166
 167     s->mse += mse;
 168     s->nb_frames++;
 169
 170     for (j = 0; j < s->nb_components; j++) {
 171         c = s->is_rgb ? s->rgba_map[j] : j;
 172         set_meta(metadata, "lavfi.psnr.mse.", s->comps[j], comp_mse[c]);
 173         set_meta(metadata, "lavfi.psnr.mse_avg", 0, mse);
 174         set_meta(metadata, "lavfi.psnr.psnr.", s->comps[j], get_psnr(comp_mse[c], 1, s->max[c]));
 175         set_meta(metadata, "lavfi.psnr.psnr_avg", 0, get_psnr(mse, 1, s->average_max));
 176     }
 177
 178     if (s->stats_file) {
 179         fprintf(s->stats_file, "n:%"PRId64" mse_avg:%0.2f ", s->nb_frames, mse);
 180         for (j = 0; j < s->nb_components; j++) {
 181             c = s->is_rgb ? s->rgba_map[j] : j;
 182             fprintf(s->stats_file, "mse_%c:%0.2f ", s->comps[j], comp_mse[c]);
 183         }
 184         for (j = 0; j < s->nb_components; j++) {
 185             c = s->is_rgb ? s->rgba_map[j] : j;
 186             fprintf(s->stats_file, "psnr_%c:%0.2f ", s->comps[j],
 187                     get_psnr(comp_mse[c], 1, s->max[c]));
 188         }
 189         fprintf(s->stats_file, "\n");
 190     }
 191
 192     return main;
 193 }
 194
 195 static av_cold int init(AVFilterContext *ctx)
 196 {
 197     PSNRContext *s = ctx->priv;
 198
 199     s->min_mse = +INFINITY;
 200     s->max_mse = -INFINITY;
 201
 202     if (s->stats_file_str) {
 203         s->stats_file = fopen(s->stats_file_str, "w");
 204         if (!s->stats_file) {
 205             int err = AVERROR(errno);
 206             char buf[128];
 207             av_strerror(err, buf, sizeof(buf));
 208             av_log(ctx, AV_LOG_ERROR, "Could not open stats file %s: %s\n",
 209                    s->stats_file_str, buf);
 210             return err;
 211         }
 212     }
 213
 214     s->dinput.process = do_psnr;
 215     return 0;
 216 }
 217
 218 static int query_formats(AVFilterContext *ctx)
 219 {
 220     static const enum PixelFormat pix_fmts[] = {
 221         AV_PIX_FMT_GRAY8, AV_PIX_FMT_GRAY16,
 222 #define PF_NOALPHA(suf) AV_PIX_FMT_YUV420##suf,  AV_PIX_FMT_YUV422##suf,  AV_PIX_FMT_YUV444##suf
 223 #define PF_ALPHA(suf)   AV_PIX_FMT_YUVA420##suf, AV_PIX_FMT_YUVA422##suf, AV_PIX_FMT_YUVA444##suf
 224 #define PF(suf)         PF_NOALPHA(suf), PF_ALPHA(suf)
 225         PF(P), PF(P9), PF(P10), PF_NOALPHA(P12), PF_NOALPHA(P14), PF(P16),
 226         AV_PIX_FMT_YUV440P, AV_PIX_FMT_YUV411P, AV_PIX_FMT_YUV410P,
 227         AV_PIX_FMT_YUVJ411P, AV_PIX_FMT_YUVJ420P, AV_PIX_FMT_YUVJ422P,
 228         AV_PIX_FMT_YUVJ440P, AV_PIX_FMT_YUVJ444P,
 229         AV_PIX_FMT_GBRP, AV_PIX_FMT_GBRP9, AV_PIX_FMT_GBRP10,
 230         AV_PIX_FMT_GBRP12, AV_PIX_FMT_GBRP14, AV_PIX_FMT_GBRP16,
 231         AV_PIX_FMT_GBRAP, AV_PIX_FMT_GBRAP16,
 232         AV_PIX_FMT_NONE
 233     };
 234
 235     ff_set_common_formats(ctx, ff_make_format_list(pix_fmts));
 236     return 0;
 237 }
 238
 239 static int config_input_ref(AVFilterLink *inlink)
 240 {
 241     const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(inlink->format);
 242     AVFilterContext *ctx  = inlink->dst;
 243     PSNRContext *s = ctx->priv;
 244     int j;
 245
 246     s->nb_components = desc->nb_components;
 247     if (ctx->inputs[0]->w != ctx->inputs[1]->w ||
 248         ctx->inputs[0]->h != ctx->inputs[1]->h) {
 249         av_log(ctx, AV_LOG_ERROR, "Width and height of input videos must be same.\n");
 250         return AVERROR(EINVAL);
 251     }
 252     if (ctx->inputs[0]->format != ctx->inputs[1]->format) {
 253         av_log(ctx, AV_LOG_ERROR, "Inputs must be of same pixel format.\n");
 254         return AVERROR(EINVAL);
 255     }
 256
 257     switch (inlink->format) {
 258     case AV_PIX_FMT_GRAY8:
 259     case AV_PIX_FMT_GRAY16:
 260     case AV_PIX_FMT_GBRP:
 261     case AV_PIX_FMT_GBRP9:
 262     case AV_PIX_FMT_GBRP10:
 263     case AV_PIX_FMT_GBRP12:
 264     case AV_PIX_FMT_GBRP14:
 265     case AV_PIX_FMT_GBRP16:
 266     case AV_PIX_FMT_GBRAP:
 267     case AV_PIX_FMT_GBRAP16:
 268     case AV_PIX_FMT_YUVJ411P:
 269     case AV_PIX_FMT_YUVJ420P:
 270     case AV_PIX_FMT_YUVJ422P:
 271     case AV_PIX_FMT_YUVJ440P:
 272     case AV_PIX_FMT_YUVJ444P:
 273         s->max[0] = (1 << (desc->comp[0].depth_minus1 + 1)) - 1;
 274         s->max[1] = (1 << (desc->comp[1].depth_minus1 + 1)) - 1;
 275         s->max[2] = (1 << (desc->comp[2].depth_minus1 + 1)) - 1;
 276         s->max[3] = (1 << (desc->comp[3].depth_minus1 + 1)) - 1;
 277         break;
 278     default:
 279         s->max[0] = 235 * (1 << (desc->comp[0].depth_minus1 - 7));
 280         s->max[1] = 240 * (1 << (desc->comp[1].depth_minus1 - 7));
 281         s->max[2] = 240 * (1 << (desc->comp[2].depth_minus1 - 7));
 282         s->max[3] = (1 << (desc->comp[3].depth_minus1 + 1)) - 1;
 283     }
 284
 285     s->is_rgb = ff_fill_rgba_map(s->rgba_map, inlink->format) >= 0;
 286     s->comps[0] = s->is_rgb ? 'r' : 'y' ;
 287     s->comps[1] = s->is_rgb ? 'g' : 'u' ;
 288     s->comps[2] = s->is_rgb ? 'b' : 'v' ;
 289     s->comps[3] = 'a';
 290
 291     for (j = 0; j < s->nb_components; j++)
 292         s->average_max += s->max[j];
 293     s->average_max /= s->nb_components;
 294
 295     s->planeheight[1] = s->planeheight[2] = FF_CEIL_RSHIFT(inlink->h, desc->log2_chroma_h);
 296     s->planeheight[0] = s->planeheight[3] = inlink->h;
 297     s->planewidth[1]  = s->planewidth[2]  = FF_CEIL_RSHIFT(inlink->w, desc->log2_chroma_w);
 298     s->planewidth[0]  = s->planewidth[3]  = inlink->w;
 299
 300     s->compute_mse = desc->comp[0].depth_minus1 > 7 ? compute_images_mse_16bit : compute_images_mse;
 301
 302     return 0;
 303 }
 304
 305 static int config_output(AVFilterLink *outlink)
 306 {
 307     AVFilterContext *ctx = outlink->src;
 308     PSNRContext *s = ctx->priv;
 309     AVFilterLink *mainlink = ctx->inputs[0];
 310     int ret;
 311
 312     outlink->w = mainlink->w;
 313     outlink->h = mainlink->h;
 314     outlink->time_base = mainlink->time_base;
 315     outlink->sample_aspect_ratio = mainlink->sample_aspect_ratio;
 316     outlink->frame_rate = mainlink->frame_rate;
 317     if ((ret = ff_dualinput_init(ctx, &s->dinput)) < 0)
 318         return ret;
 319
 320     return 0;
 321 }
 322
 323 static int filter_frame(AVFilterLink *inlink, AVFrame *inpicref)
 324 {
 325     PSNRContext *s = inlink->dst->priv;
 326     return ff_dualinput_filter_frame(&s->dinput, inlink, inpicref);
 327 }
 328
 329 static int request_frame(AVFilterLink *outlink)
 330 {
 331     PSNRContext *s = outlink->src->priv;
 332     return ff_dualinput_request_frame(&s->dinput, outlink);
 333 }
 334
 335 static av_cold void uninit(AVFilterContext *ctx)
 336 {
 337     PSNRContext *s = ctx->priv;
 338
 339     if (s->nb_frames > 0) {
 340         av_log(ctx, AV_LOG_INFO, "PSNR average:%0.2f min:%0.2f max:%0.2f\n",
 341                get_psnr(s->mse, s->nb_frames, s->average_max),
 342                get_psnr(s->max_mse, 1, s->average_max),
 343                get_psnr(s->min_mse, 1, s->average_max));
 344     }
 345
 346     ff_dualinput_uninit(&s->dinput);
 347
 348     if (s->stats_file)
 349         fclose(s->stats_file);
 350 }
 351
 352 static const AVFilterPad psnr_inputs[] = {
 353     {
 354         .name         = "main",
 355         .type         = AVMEDIA_TYPE_VIDEO,
 356         .filter_frame = filter_frame,
 357     },{
 358         .name         = "reference",
 359         .type         = AVMEDIA_TYPE_VIDEO,
 360         .filter_frame = filter_frame,
 361         .config_props = config_input_ref,
 362     },
 363     { NULL }
 364 };
 365
 366 static const AVFilterPad psnr_outputs[] = {
 367     {
 368         .name          = "default",
 369         .type          = AVMEDIA_TYPE_VIDEO,
 370         .config_props  = config_output,
 371         .request_frame = request_frame,
 372     },
 373     { NULL }
 374 };
 375
 376 AVFilter ff_vf_psnr = {
 377     .name          = "psnr",
 378     .description   = NULL_IF_CONFIG_SMALL("Calculate the PSNR between two video streams."),
 379     .init          = init,
 380     .uninit        = uninit,
 381     .query_formats = query_formats,
 382     .priv_size     = sizeof(PSNRContext),
 383     .priv_class    = &psnr_class,
 384     .inputs        = psnr_inputs,
 385     .outputs       = psnr_outputs,
 386 };