FFmpeg
 All Data Structures Namespaces Files Functions Variables Typedefs Enumerations Enumerator Macros Groups Pages
vf_signature.c
Go to the documentation of this file.
1 /*
2  * Copyright (c) 2017 Gerion Entrup
3  *
4  * This file is part of FFmpeg.
5  *
6  * FFmpeg is free software; you can redistribute it and/or modify
7  * it under the terms of the GNU General Public License as published by
8  * the Free Software Foundation; either version 2 of the License, or
9  * (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
14  * GNU General Public License for more details.
15  *
16  * You should have received a copy of the GNU General Public License along
17  * with FFmpeg; if not, write to the Free Software Foundation, Inc.,
18  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
19  */
20 
21 /**
22  * @file
23  * MPEG-7 video signature calculation and lookup filter
24  * @see http://epubs.surrey.ac.uk/531590/1/MPEG-7%20Video%20Signature%20Author%27s%20Copy.pdf
25  */
26 
27 #include <float.h>
28 #include "libavcodec/put_bits.h"
29 #include "libavformat/avformat.h"
30 #include "libavutil/opt.h"
31 #include "libavutil/avstring.h"
32 #include "libavutil/intreadwrite.h"
33 #include "libavutil/timestamp.h"
34 #include "avfilter.h"
35 #include "internal.h"
36 #include "signature.h"
37 #include "signature_lookup.c"
38 
39 #define OFFSET(x) offsetof(SignatureContext, x)
40 #define FLAGS AV_OPT_FLAG_FILTERING_PARAM | AV_OPT_FLAG_VIDEO_PARAM
41 #define BLOCK_LCM (int64_t) 476985600
42 
43 static const AVOption signature_options[] = {
44  { "detectmode", "set the detectmode",
45  OFFSET(mode), AV_OPT_TYPE_INT, {.i64 = MODE_OFF}, 0, NB_LOOKUP_MODE-1, FLAGS, "mode" },
46  { "off", NULL, 0, AV_OPT_TYPE_CONST, {.i64 = MODE_OFF}, 0, 0, .flags = FLAGS, "mode" },
47  { "full", NULL, 0, AV_OPT_TYPE_CONST, {.i64 = MODE_FULL}, 0, 0, .flags = FLAGS, "mode" },
48  { "fast", NULL, 0, AV_OPT_TYPE_CONST, {.i64 = MODE_FAST}, 0, 0, .flags = FLAGS, "mode" },
49  { "nb_inputs", "number of inputs",
50  OFFSET(nb_inputs), AV_OPT_TYPE_INT, {.i64 = 1}, 1, INT_MAX, FLAGS },
51  { "filename", "filename for output files",
52  OFFSET(filename), AV_OPT_TYPE_STRING, {.str = ""}, 0, NB_FORMATS-1, FLAGS },
53  { "format", "set output format",
54  OFFSET(format), AV_OPT_TYPE_INT, {.i64 = FORMAT_BINARY}, 0, 1, FLAGS , "format" },
55  { "binary", 0, 0, AV_OPT_TYPE_CONST, {.i64=FORMAT_BINARY}, 0, 0, FLAGS, "format" },
56  { "xml", 0, 0, AV_OPT_TYPE_CONST, {.i64=FORMAT_XML}, 0, 0, FLAGS, "format" },
57  { "th_d", "threshold to detect one word as similar",
58  OFFSET(thworddist), AV_OPT_TYPE_INT, {.i64 = 9000}, 1, INT_MAX, FLAGS },
59  { "th_dc", "threshold to detect all words as similar",
60  OFFSET(thcomposdist), AV_OPT_TYPE_INT, {.i64 = 60000}, 1, INT_MAX, FLAGS },
61  { "th_xh", "threshold to detect frames as similar",
62  OFFSET(thl1), AV_OPT_TYPE_INT, {.i64 = 116}, 1, INT_MAX, FLAGS },
63  { "th_di", "minimum length of matching sequence in frames",
64  OFFSET(thdi), AV_OPT_TYPE_INT, {.i64 = 0}, 0, INT_MAX, FLAGS },
65  { "th_it", "threshold for relation of good to all frames",
66  OFFSET(thit), AV_OPT_TYPE_DOUBLE, {.dbl = 0.5}, 0.0, 1.0, FLAGS },
67  { NULL }
68 };
69 
71 
73 {
74  /* all formats with a separate gray value */
75  static const enum AVPixelFormat pix_fmts[] = {
85  };
86 
87  return ff_set_common_formats(ctx, ff_make_format_list(pix_fmts));
88 }
89 
90 static int config_input(AVFilterLink *inlink)
91 {
92  AVFilterContext *ctx = inlink->dst;
93  SignatureContext *sic = ctx->priv;
94  StreamContext *sc = &(sic->streamcontexts[FF_INLINK_IDX(inlink)]);
95 
96  sc->time_base = inlink->time_base;
97  /* test for overflow */
98  sc->divide = (((uint64_t) inlink->w/32) * (inlink->w/32 + 1) * (inlink->h/32 * inlink->h/32 + 1) > INT64_MAX / (BLOCK_LCM * 255));
99  if (sc->divide) {
100  av_log(ctx, AV_LOG_WARNING, "Input dimension too high for precise calculation, numbers will be rounded.\n");
101  }
102  sc->w = inlink->w;
103  sc->h = inlink->h;
104  return 0;
105 }
106 
107 static int get_block_size(const Block *b)
108 {
109  return (b->to.y - b->up.y + 1) * (b->to.x - b->up.x + 1);
110 }
111 
112 static uint64_t get_block_sum(StreamContext *sc, uint64_t intpic[32][32], const Block *b)
113 {
114  uint64_t sum = 0;
115 
116  int x0, y0, x1, y1;
117 
118  x0 = b->up.x;
119  y0 = b->up.y;
120  x1 = b->to.x;
121  y1 = b->to.y;
122 
123  if (x0-1 >= 0 && y0-1 >= 0) {
124  sum = intpic[y1][x1] + intpic[y0-1][x0-1] - intpic[y1][x0-1] - intpic[y0-1][x1];
125  } else if (x0-1 >= 0) {
126  sum = intpic[y1][x1] - intpic[y1][x0-1];
127  } else if (y0-1 >= 0) {
128  sum = intpic[y1][x1] - intpic[y0-1][x1];
129  } else {
130  sum = intpic[y1][x1];
131  }
132  return sum;
133 }
134 
135 static int cmp(const uint64_t *a, const uint64_t *b)
136 {
137  return *a < *b ? -1 : ( *a > *b ? 1 : 0 );
138 }
139 
140 /**
141  * sets the bit at position pos to 1 in data
142  */
143 static void set_bit(uint8_t* data, size_t pos)
144 {
145  uint8_t mask = 1 << 7-(pos%8);
146  data[pos/8] |= mask;
147 }
148 
149 static int filter_frame(AVFilterLink *inlink, AVFrame *picref)
150 {
151  AVFilterContext *ctx = inlink->dst;
152  SignatureContext *sic = ctx->priv;
153  StreamContext *sc = &(sic->streamcontexts[FF_INLINK_IDX(inlink)]);
154  FineSignature* fs;
155 
156  static const uint8_t pot3[5] = { 3*3*3*3, 3*3*3, 3*3, 3, 1 };
157  /* indexes of words : 210,217,219,274,334 44,175,233,270,273 57,70,103,237,269 100,285,295,337,354 101,102,111,275,296
158  s2usw = sorted to unsorted wordvec: 44 is at index 5, 57 at index 10...
159  */
160  static const unsigned int wordvec[25] = {44,57,70,100,101,102,103,111,175,210,217,219,233,237,269,270,273,274,275,285,295,296,334,337,354};
161  static const uint8_t s2usw[25] = { 5,10,11, 15, 20, 21, 12, 22, 6, 0, 1, 2, 7, 13, 14, 8, 9, 3, 23, 16, 17, 24, 4, 18, 19};
162 
163  uint8_t wordt2b[5] = { 0, 0, 0, 0, 0 }; /* word ternary to binary */
164  uint64_t intpic[32][32];
165  uint64_t rowcount;
166  uint8_t *p = picref->data[0];
167  int inti, intj;
168  int *intjlut;
169 
170  uint64_t conflist[DIFFELEM_SIZE];
171  int f = 0, g = 0, w = 0;
172  int32_t dh1 = 1, dh2 = 1, dw1 = 1, dw2 = 1, a, b;
173  int64_t denom;
174  int i, j, k, ternary;
175  uint64_t blocksum;
176  int blocksize;
177  int64_t th; /* threshold */
178  int64_t sum;
179 
180  int64_t precfactor = (sc->divide) ? 65536 : BLOCK_LCM;
181 
182  /* initialize fs */
183  if (sc->curfinesig) {
184  fs = av_mallocz(sizeof(FineSignature));
185  if (!fs)
186  return AVERROR(ENOMEM);
187  sc->curfinesig->next = fs;
188  fs->prev = sc->curfinesig;
189  sc->curfinesig = fs;
190  } else {
191  fs = sc->curfinesig = sc->finesiglist;
192  sc->curcoarsesig1->first = fs;
193  }
194 
195  fs->pts = picref->pts;
196  fs->index = sc->lastindex++;
197 
198  memset(intpic, 0, sizeof(uint64_t)*32*32);
199  intjlut = av_malloc_array(inlink->w, sizeof(int));
200  if (!intjlut)
201  return AVERROR(ENOMEM);
202  for (i = 0; i < inlink->w; i++) {
203  intjlut[i] = (i*32)/inlink->w;
204  }
205 
206  for (i = 0; i < inlink->h; i++) {
207  inti = (i*32)/inlink->h;
208  for (j = 0; j < inlink->w; j++) {
209  intj = intjlut[j];
210  intpic[inti][intj] += p[j];
211  }
212  p += picref->linesize[0];
213  }
214  av_freep(&intjlut);
215 
216  /* The following calculates a summed area table (intpic) and brings the numbers
217  * in intpic to the same denominator.
218  * So you only have to handle the numinator in the following sections.
219  */
220  dh1 = inlink->h / 32;
221  if (inlink->h % 32)
222  dh2 = dh1 + 1;
223  dw1 = inlink->w / 32;
224  if (inlink->w % 32)
225  dw2 = dw1 + 1;
226  denom = (sc->divide) ? dh1 * dh2 * dw1 * dw2 : 1;
227 
228  for (i = 0; i < 32; i++) {
229  rowcount = 0;
230  a = 1;
231  if (dh2 > 1) {
232  a = ((inlink->h*(i+1))%32 == 0) ? (inlink->h*(i+1))/32 - 1 : (inlink->h*(i+1))/32;
233  a -= ((inlink->h*i)%32 == 0) ? (inlink->h*i)/32 - 1 : (inlink->h*i)/32;
234  a = (a == dh1)? dh2 : dh1;
235  }
236  for (j = 0; j < 32; j++) {
237  b = 1;
238  if (dw2 > 1) {
239  b = ((inlink->w*(j+1))%32 == 0) ? (inlink->w*(j+1))/32 - 1 : (inlink->w*(j+1))/32;
240  b -= ((inlink->w*j)%32 == 0) ? (inlink->w*j)/32 - 1 : (inlink->w*j)/32;
241  b = (b == dw1)? dw2 : dw1;
242  }
243  rowcount += intpic[i][j] * a * b * precfactor / denom;
244  if (i > 0) {
245  intpic[i][j] = intpic[i-1][j] + rowcount;
246  } else {
247  intpic[i][j] = rowcount;
248  }
249  }
250  }
251 
252  denom = (sc->divide) ? 1 : dh1 * dh2 * dw1 * dw2;
253 
254  for (i = 0; i < ELEMENT_COUNT; i++) {
255  const ElemCat* elemcat = elements[i];
256  int64_t* elemsignature;
257  uint64_t* sortsignature;
258 
259  elemsignature = av_malloc_array(elemcat->elem_count, sizeof(int64_t));
260  if (!elemsignature)
261  return AVERROR(ENOMEM);
262  sortsignature = av_malloc_array(elemcat->elem_count, sizeof(int64_t));
263  if (!sortsignature) {
264  av_freep(&elemsignature);
265  return AVERROR(ENOMEM);
266  }
267 
268  for (j = 0; j < elemcat->elem_count; j++) {
269  blocksum = 0;
270  blocksize = 0;
271  for (k = 0; k < elemcat->left_count; k++) {
272  blocksum += get_block_sum(sc, intpic, &elemcat->blocks[j*elemcat->block_count+k]);
273  blocksize += get_block_size(&elemcat->blocks[j*elemcat->block_count+k]);
274  }
275  sum = blocksum / blocksize;
276  if (elemcat->av_elem) {
277  sum -= 128 * precfactor * denom;
278  } else {
279  blocksum = 0;
280  blocksize = 0;
281  for (; k < elemcat->block_count; k++) {
282  blocksum += get_block_sum(sc, intpic, &elemcat->blocks[j*elemcat->block_count+k]);
283  blocksize += get_block_size(&elemcat->blocks[j*elemcat->block_count+k]);
284  }
285  sum -= blocksum / blocksize;
286  conflist[g++] = FFABS(sum * 8 / (precfactor * denom));
287  }
288 
289  elemsignature[j] = sum;
290  sortsignature[j] = FFABS(sum);
291  }
292 
293  /* get threshold */
294  qsort(sortsignature, elemcat->elem_count, sizeof(uint64_t), (void*) cmp);
295  th = sortsignature[(int) (elemcat->elem_count*0.333)];
296 
297  /* ternarize */
298  for (j = 0; j < elemcat->elem_count; j++) {
299  if (elemsignature[j] < -th) {
300  ternary = 0;
301  } else if (elemsignature[j] <= th) {
302  ternary = 1;
303  } else {
304  ternary = 2;
305  }
306  fs->framesig[f/5] += ternary * pot3[f%5];
307 
308  if (f == wordvec[w]) {
309  fs->words[s2usw[w]/5] += ternary * pot3[wordt2b[s2usw[w]/5]++];
310  if (w < 24)
311  w++;
312  }
313  f++;
314  }
315  av_freep(&elemsignature);
316  av_freep(&sortsignature);
317  }
318 
319  /* confidence */
320  qsort(conflist, DIFFELEM_SIZE, sizeof(uint64_t), (void*) cmp);
321  fs->confidence = FFMIN(conflist[DIFFELEM_SIZE/2], 255);
322 
323  /* coarsesignature */
324  if (sc->coarsecount == 0) {
325  if (sc->curcoarsesig2) {
327  if (!sc->curcoarsesig1)
328  return AVERROR(ENOMEM);
329  sc->curcoarsesig1->first = fs;
330  sc->curcoarsesig2->next = sc->curcoarsesig1;
331  sc->coarseend = sc->curcoarsesig1;
332  }
333  }
334  if (sc->coarsecount == 45) {
335  sc->midcoarse = 1;
337  if (!sc->curcoarsesig2)
338  return AVERROR(ENOMEM);
339  sc->curcoarsesig2->first = fs;
340  sc->curcoarsesig1->next = sc->curcoarsesig2;
341  sc->coarseend = sc->curcoarsesig2;
342  }
343  for (i = 0; i < 5; i++) {
344  set_bit(sc->curcoarsesig1->data[i], fs->words[i]);
345  }
346  /* assuming the actual frame is the last */
347  sc->curcoarsesig1->last = fs;
348  if (sc->midcoarse) {
349  for (i = 0; i < 5; i++) {
350  set_bit(sc->curcoarsesig2->data[i], fs->words[i]);
351  }
352  sc->curcoarsesig2->last = fs;
353  }
354 
355  sc->coarsecount = (sc->coarsecount+1)%90;
356 
357  /* debug printing finesignature */
358  if (av_log_get_level() == AV_LOG_DEBUG) {
359  av_log(ctx, AV_LOG_DEBUG, "input %d, confidence: %d\n", FF_INLINK_IDX(inlink), fs->confidence);
360 
361  av_log(ctx, AV_LOG_DEBUG, "words:");
362  for (i = 0; i < 5; i++) {
363  av_log(ctx, AV_LOG_DEBUG, " %d:", fs->words[i] );
364  av_log(ctx, AV_LOG_DEBUG, " %d", fs->words[i] / pot3[0] );
365  for (j = 1; j < 5; j++)
366  av_log(ctx, AV_LOG_DEBUG, ",%d", fs->words[i] % pot3[j-1] / pot3[j] );
367  av_log(ctx, AV_LOG_DEBUG, ";");
368  }
369  av_log(ctx, AV_LOG_DEBUG, "\n");
370 
371  av_log(ctx, AV_LOG_DEBUG, "framesignature:");
372  for (i = 0; i < SIGELEM_SIZE/5; i++) {
373  av_log(ctx, AV_LOG_DEBUG, " %d", fs->framesig[i] / pot3[0] );
374  for (j = 1; j < 5; j++)
375  av_log(ctx, AV_LOG_DEBUG, ",%d", fs->framesig[i] % pot3[j-1] / pot3[j] );
376  }
377  av_log(ctx, AV_LOG_DEBUG, "\n");
378  }
379 
380  if (FF_INLINK_IDX(inlink) == 0)
381  return ff_filter_frame(inlink->dst->outputs[0], picref);
382  return 1;
383 }
384 
385 static int xml_export(AVFilterContext *ctx, StreamContext *sc, const char* filename)
386 {
387  FineSignature* fs;
388  CoarseSignature* cs;
389  int i, j;
390  FILE* f;
391  unsigned int pot3[5] = { 3*3*3*3, 3*3*3, 3*3, 3, 1 };
392 
393  f = fopen(filename, "w");
394  if (!f) {
395  int err = AVERROR(EINVAL);
396  char buf[128];
397  av_strerror(err, buf, sizeof(buf));
398  av_log(ctx, AV_LOG_ERROR, "cannot open xml file %s: %s\n", filename, buf);
399  return err;
400  }
401 
402  /* header */
403  fprintf(f, "<?xml version='1.0' encoding='ASCII' ?>\n");
404  fprintf(f, "<Mpeg7 xmlns=\"urn:mpeg:mpeg7:schema:2001\" xmlns:xsi=\"http://www.w3.org/2001/XMLSchema-instance\" xsi:schemaLocation=\"urn:mpeg:mpeg7:schema:2001 schema/Mpeg7-2001.xsd\">\n");
405  fprintf(f, " <DescriptionUnit xsi:type=\"DescriptorCollectionType\">\n");
406  fprintf(f, " <Descriptor xsi:type=\"VideoSignatureType\">\n");
407  fprintf(f, " <VideoSignatureRegion>\n");
408  fprintf(f, " <VideoSignatureSpatialRegion>\n");
409  fprintf(f, " <Pixel>0 0 </Pixel>\n");
410  fprintf(f, " <Pixel>%d %d </Pixel>\n", sc->w - 1, sc->h - 1);
411  fprintf(f, " </VideoSignatureSpatialRegion>\n");
412  fprintf(f, " <StartFrameOfSpatialRegion>0</StartFrameOfSpatialRegion>\n");
413  /* hoping num is 1, other values are vague */
414  fprintf(f, " <MediaTimeUnit>%d</MediaTimeUnit>\n", sc->time_base.den / sc->time_base.num);
415  fprintf(f, " <MediaTimeOfSpatialRegion>\n");
416  fprintf(f, " <StartMediaTimeOfSpatialRegion>0</StartMediaTimeOfSpatialRegion>\n");
417  fprintf(f, " <EndMediaTimeOfSpatialRegion>%" PRIu64 "</EndMediaTimeOfSpatialRegion>\n", sc->coarseend->last->pts);
418  fprintf(f, " </MediaTimeOfSpatialRegion>\n");
419 
420  /* coarsesignatures */
421  for (cs = sc->coarsesiglist; cs; cs = cs->next) {
422  fprintf(f, " <VSVideoSegment>\n");
423  fprintf(f, " <StartFrameOfSegment>%" PRIu32 "</StartFrameOfSegment>\n", cs->first->index);
424  fprintf(f, " <EndFrameOfSegment>%" PRIu32 "</EndFrameOfSegment>\n", cs->last->index);
425  fprintf(f, " <MediaTimeOfSegment>\n");
426  fprintf(f, " <StartMediaTimeOfSegment>%" PRIu64 "</StartMediaTimeOfSegment>\n", cs->first->pts);
427  fprintf(f, " <EndMediaTimeOfSegment>%" PRIu64 "</EndMediaTimeOfSegment>\n", cs->last->pts);
428  fprintf(f, " </MediaTimeOfSegment>\n");
429  for (i = 0; i < 5; i++) {
430  fprintf(f, " <BagOfWords>");
431  for (j = 0; j < 31; j++) {
432  uint8_t n = cs->data[i][j];
433  if (j < 30) {
434  fprintf(f, "%d %d %d %d %d %d %d %d ", (n & 0x80) >> 7,
435  (n & 0x40) >> 6,
436  (n & 0x20) >> 5,
437  (n & 0x10) >> 4,
438  (n & 0x08) >> 3,
439  (n & 0x04) >> 2,
440  (n & 0x02) >> 1,
441  (n & 0x01));
442  } else {
443  /* print only 3 bit in last byte */
444  fprintf(f, "%d %d %d ", (n & 0x80) >> 7,
445  (n & 0x40) >> 6,
446  (n & 0x20) >> 5);
447  }
448  }
449  fprintf(f, "</BagOfWords>\n");
450  }
451  fprintf(f, " </VSVideoSegment>\n");
452  }
453 
454  /* finesignatures */
455  for (fs = sc->finesiglist; fs; fs = fs->next) {
456  fprintf(f, " <VideoFrame>\n");
457  fprintf(f, " <MediaTimeOfFrame>%" PRIu64 "</MediaTimeOfFrame>\n", fs->pts);
458  /* confidence */
459  fprintf(f, " <FrameConfidence>%d</FrameConfidence>\n", fs->confidence);
460  /* words */
461  fprintf(f, " <Word>");
462  for (i = 0; i < 5; i++) {
463  fprintf(f, "%d ", fs->words[i]);
464  if (i < 4) {
465  fprintf(f, " ");
466  }
467  }
468  fprintf(f, "</Word>\n");
469  /* framesignature */
470  fprintf(f, " <FrameSignature>");
471  for (i = 0; i< SIGELEM_SIZE/5; i++) {
472  if (i > 0) {
473  fprintf(f, " ");
474  }
475  fprintf(f, "%d ", fs->framesig[i] / pot3[0]);
476  for (j = 1; j < 5; j++)
477  fprintf(f, " %d ", fs->framesig[i] % pot3[j-1] / pot3[j] );
478  }
479  fprintf(f, "</FrameSignature>\n");
480  fprintf(f, " </VideoFrame>\n");
481  }
482  fprintf(f, " </VideoSignatureRegion>\n");
483  fprintf(f, " </Descriptor>\n");
484  fprintf(f, " </DescriptionUnit>\n");
485  fprintf(f, "</Mpeg7>\n");
486 
487  fclose(f);
488  return 0;
489 }
490 
491 static int binary_export(AVFilterContext *ctx, StreamContext *sc, const char* filename)
492 {
493  FILE* f;
494  FineSignature* fs;
495  CoarseSignature* cs;
496  uint32_t numofsegments = (sc->lastindex + 44)/45;
497  int i, j;
499  /* buffer + header + coarsesignatures + finesignature */
500  int len = (512 + 6 * 32 + 3*16 + 2 +
501  numofsegments * (4*32 + 1 + 5*243) +
502  sc->lastindex * (2 + 32 + 6*8 + 608)) / 8;
503  uint8_t* buffer = av_malloc_array(len, sizeof(uint8_t));
504  if (!buffer)
505  return AVERROR(ENOMEM);
506 
507  f = fopen(filename, "wb");
508  if (!f) {
509  int err = AVERROR(EINVAL);
510  char buf[128];
511  av_strerror(err, buf, sizeof(buf));
512  av_log(ctx, AV_LOG_ERROR, "cannot open file %s: %s\n", filename, buf);
513  av_freep(&buffer);
514  return err;
515  }
516  init_put_bits(&buf, buffer, len);
517 
518  put_bits32(&buf, 1); /* NumOfSpatial Regions, only 1 supported */
519  put_bits(&buf, 1, 1); /* SpatialLocationFlag, always the whole image */
520  put_bits32(&buf, 0); /* PixelX,1 PixelY,1, 0,0 */
521  put_bits(&buf, 16, sc->w-1 & 0xFFFF); /* PixelX,2 */
522  put_bits(&buf, 16, sc->h-1 & 0xFFFF); /* PixelY,2 */
523  put_bits32(&buf, 0); /* StartFrameOfSpatialRegion */
524  put_bits32(&buf, sc->lastindex); /* NumOfFrames */
525  /* hoping num is 1, other values are vague */
526  /* den/num might be greater than 16 bit, so cutting it */
527  put_bits(&buf, 16, 0xFFFF & (sc->time_base.den / sc->time_base.num)); /* MediaTimeUnit */
528  put_bits(&buf, 1, 1); /* MediaTimeFlagOfSpatialRegion */
529  put_bits32(&buf, 0); /* StartMediaTimeOfSpatialRegion */
530  put_bits32(&buf, 0xFFFFFFFF & sc->coarseend->last->pts); /* EndMediaTimeOfSpatialRegion */
531  put_bits32(&buf, numofsegments); /* NumOfSegments */
532  /* coarsesignatures */
533  for (cs = sc->coarsesiglist; cs; cs = cs->next) {
534  put_bits32(&buf, cs->first->index); /* StartFrameOfSegment */
535  put_bits32(&buf, cs->last->index); /* EndFrameOfSegment */
536  put_bits(&buf, 1, 1); /* MediaTimeFlagOfSegment */
537  put_bits32(&buf, 0xFFFFFFFF & cs->first->pts); /* StartMediaTimeOfSegment */
538  put_bits32(&buf, 0xFFFFFFFF & cs->last->pts); /* EndMediaTimeOfSegment */
539  for (i = 0; i < 5; i++) {
540  /* put 243 bits ( = 7 * 32 + 19 = 8 * 28 + 19) into buffer */
541  for (j = 0; j < 30; j++) {
542  put_bits(&buf, 8, cs->data[i][j]);
543  }
544  put_bits(&buf, 3, cs->data[i][30] >> 5);
545  }
546  }
547  /* finesignatures */
548  put_bits(&buf, 1, 0); /* CompressionFlag, only 0 supported */
549  for (fs = sc->finesiglist; fs; fs = fs->next) {
550  put_bits(&buf, 1, 1); /* MediaTimeFlagOfFrame */
551  put_bits32(&buf, 0xFFFFFFFF & fs->pts); /* MediaTimeOfFrame */
552  put_bits(&buf, 8, fs->confidence); /* FrameConfidence */
553  for (i = 0; i < 5; i++) {
554  put_bits(&buf, 8, fs->words[i]); /* Words */
555  }
556  /* framesignature */
557  for (i = 0; i < SIGELEM_SIZE/5; i++) {
558  put_bits(&buf, 8, fs->framesig[i]);
559  }
560  }
561 
562  avpriv_align_put_bits(&buf);
563  flush_put_bits(&buf);
564  fwrite(buffer, 1, put_bits_count(&buf)/8, f);
565  fclose(f);
566  av_freep(&buffer);
567  return 0;
568 }
569 
570 static int export(AVFilterContext *ctx, StreamContext *sc, int input)
571 {
572  SignatureContext* sic = ctx->priv;
573  char filename[1024];
574 
575  if (sic->nb_inputs > 1) {
576  /* error already handled */
577  av_assert0(av_get_frame_filename(filename, sizeof(filename), sic->filename, input) == 0);
578  } else {
579  strcpy(filename, sic->filename);
580  }
581  if (sic->format == FORMAT_XML) {
582  return xml_export(ctx, sc, filename);
583  } else {
584  return binary_export(ctx, sc, filename);
585  }
586 }
587 
588 static int request_frame(AVFilterLink *outlink)
589 {
590  AVFilterContext *ctx = outlink->src;
591  SignatureContext *sic = ctx->priv;
592  StreamContext *sc, *sc2;
593  MatchingInfo match;
594  int i, j, ret;
595  int lookup = 1; /* indicates wheather EOF of all files is reached */
596 
597  /* process all inputs */
598  for (i = 0; i < sic->nb_inputs; i++){
599  sc = &(sic->streamcontexts[i]);
600 
601  ret = ff_request_frame(ctx->inputs[i]);
602 
603  /* return if unexpected error occurs in input stream */
604  if (ret < 0 && ret != AVERROR_EOF)
605  return ret;
606 
607  /* export signature at EOF */
608  if (ret == AVERROR_EOF && !sc->exported) {
609  /* export if wanted */
610  if (strlen(sic->filename) > 0) {
611  if (export(ctx, sc, i) < 0)
612  return ret;
613  }
614  sc->exported = 1;
615  }
616  lookup &= sc->exported;
617  }
618 
619  /* signature lookup */
620  if (lookup && sic->mode != MODE_OFF) {
621  /* iterate over every pair */
622  for (i = 0; i < sic->nb_inputs; i++) {
623  sc = &(sic->streamcontexts[i]);
624  for (j = i+1; j < sic->nb_inputs; j++) {
625  sc2 = &(sic->streamcontexts[j]);
626  match = lookup_signatures(ctx, sic, sc, sc2, sic->mode);
627  if (match.score != 0) {
628  av_log(ctx, AV_LOG_INFO, "matching of video %d at %f and %d at %f, %d frames matching\n",
629  i, ((double) match.first->pts * sc->time_base.num) / sc->time_base.den,
630  j, ((double) match.second->pts * sc2->time_base.num) / sc2->time_base.den,
631  match.matchframes);
632  if (match.whole)
633  av_log(ctx, AV_LOG_INFO, "whole video matching\n");
634  } else {
635  av_log(ctx, AV_LOG_INFO, "no matching of video %d and %d\n", i, j);
636  }
637  }
638  }
639  }
640 
641  return ret;
642 }
643 
645 {
646 
647  SignatureContext *sic = ctx->priv;
648  StreamContext *sc;
649  int i, ret;
650  char tmp[1024];
651 
652  sic->streamcontexts = av_mallocz(sic->nb_inputs * sizeof(StreamContext));
653  if (!sic->streamcontexts)
654  return AVERROR(ENOMEM);
655 
656  for (i = 0; i < sic->nb_inputs; i++) {
657  AVFilterPad pad = {
659  .name = av_asprintf("in%d", i),
660  .config_props = config_input,
661  .filter_frame = filter_frame,
662  };
663 
664  if (!pad.name)
665  return AVERROR(ENOMEM);
666 
667  sc = &(sic->streamcontexts[i]);
668 
669  sc->lastindex = 0;
670  sc->finesiglist = av_mallocz(sizeof(FineSignature));
671  if (!sc->finesiglist)
672  return AVERROR(ENOMEM);
673  sc->curfinesig = NULL;
674 
676  if (!sc->coarsesiglist)
677  return AVERROR(ENOMEM);
678  sc->curcoarsesig1 = sc->coarsesiglist;
679  sc->coarseend = sc->coarsesiglist;
680  sc->coarsecount = 0;
681  sc->midcoarse = 0;
682 
683  if ((ret = ff_insert_inpad(ctx, i, &pad)) < 0) {
684  av_freep(&pad.name);
685  return ret;
686  }
687  }
688 
689  /* check filename */
690  if (sic->nb_inputs > 1 && strlen(sic->filename) > 0 && av_get_frame_filename(tmp, sizeof(tmp), sic->filename, 0) == -1) {
691  av_log(ctx, AV_LOG_ERROR, "The filename must contain %%d or %%0nd, if you have more than one input.\n");
692  return AVERROR(EINVAL);
693  }
694 
695  return 0;
696 }
697 
698 
699 
701 {
702  SignatureContext *sic = ctx->priv;
703  StreamContext *sc;
704  void* tmp;
705  FineSignature* finsig;
706  CoarseSignature* cousig;
707  int i;
708 
709 
710  /* free the lists */
711  if (sic->streamcontexts != NULL) {
712  for (i = 0; i < sic->nb_inputs; i++) {
713  sc = &(sic->streamcontexts[i]);
714  finsig = sc->finesiglist;
715  cousig = sc->coarsesiglist;
716 
717  while (finsig) {
718  tmp = finsig;
719  finsig = finsig->next;
720  av_freep(&tmp);
721  }
722  sc->finesiglist = NULL;
723 
724  while (cousig) {
725  tmp = cousig;
726  cousig = cousig->next;
727  av_freep(&tmp);
728  }
729  sc->coarsesiglist = NULL;
730  }
731  av_freep(&sic->streamcontexts);
732  }
733 }
734 
735 static int config_output(AVFilterLink *outlink)
736 {
737  AVFilterContext *ctx = outlink->src;
738  AVFilterLink *inlink = ctx->inputs[0];
739 
740  outlink->time_base = inlink->time_base;
741  outlink->frame_rate = inlink->frame_rate;
742  outlink->sample_aspect_ratio = inlink->sample_aspect_ratio;
743  outlink->w = inlink->w;
744  outlink->h = inlink->h;
745 
746  return 0;
747 }
748 
749 static const AVFilterPad signature_outputs[] = {
750  {
751  .name = "default",
752  .type = AVMEDIA_TYPE_VIDEO,
753  .request_frame = request_frame,
754  .config_props = config_output,
755  },
756  { NULL }
757 };
758 
760  .name = "signature",
761  .description = NULL_IF_CONFIG_SMALL("Calculate the MPEG-7 video signature"),
762  .priv_size = sizeof(SignatureContext),
763  .priv_class = &signature_class,
764  .init = init,
765  .uninit = uninit,
767  .outputs = signature_outputs,
768  .inputs = NULL,
770 };
static void av_unused put_bits32(PutBitContext *s, uint32_t value)
Write exactly 32 bits into a bitstream.
Definition: put_bits.h:250
#define NULL
Definition: coverity.c:32
This structure describes decoded (raw) audio or video data.
Definition: frame.h:201
int av_elem
Definition: signature.h:65
static const AVFilterPad signature_outputs[]
Definition: vf_signature.c:749
AVOption.
Definition: opt.h:246
ptrdiff_t const GLvoid * data
Definition: opengl_enc.c:101
planar YUV 4:4:4, 24bpp, (1 Cr & Cb sample per 1x1 Y samples)
Definition: pixfmt.h:67
static void put_bits(Jpeg2000EncoderContext *s, int val, int n)
put n times val bit
Definition: j2kenc.c:206
#define AV_LOG_WARNING
Something somehow does not look correct.
Definition: log.h:182
uint8_t y
Definition: signature.h:56
Main libavfilter public API header.
const char * g
Definition: vf_curves.c:112
#define OFFSET(x)
Definition: vf_signature.c:39
static int cmp(const uint64_t *a, const uint64_t *b)
Definition: vf_signature.c:135
#define AVFILTER_FLAG_DYNAMIC_INPUTS
The number of the filter inputs is not determined just by AVFilter.inputs.
Definition: avfilter.h:105
int num
Numerator.
Definition: rational.h:59
const Block * blocks
Definition: signature.h:69
const char * b
Definition: vf_curves.c:113
static void set_bit(uint8_t *data, size_t pos)
sets the bit at position pos to 1 in data
Definition: vf_signature.c:143
struct FineSignature * next
Definition: signature.h:73
void avpriv_align_put_bits(PutBitContext *s)
Pad the bitstream with zeros up to the next byte boundary.
Definition: bitstream.c:48
enum AVMediaType type
AVFilterPad type.
Definition: internal.h:65
static int filter_frame(AVFilterLink *inlink, AVFrame *picref)
Definition: vf_signature.c:149
void * av_mallocz(size_t size)
Allocate a memory block with alignment suitable for all memory accesses (including vectors if availab...
Definition: mem.c:222
StreamContext * streamcontexts
Definition: signature.h:142
uint32_t index
Definition: signature.h:76
static const AVOption signature_options[]
Definition: vf_signature.c:43
AVFilterFormats * ff_make_format_list(const int *fmts)
Create a list of supported formats.
Definition: formats.c:283
const char * name
Pad name.
Definition: internal.h:60
AVFilterLink ** inputs
array of pointers to input links
Definition: avfilter.h:346
#define av_assert0(cond)
assert() equivalent, that is always enabled.
Definition: avassert.h:37
static const char signature[]
Definition: ipmovie.c:615
int ff_filter_frame(AVFilterLink *link, AVFrame *frame)
Send a frame of data to the next filter.
Definition: avfilter.c:1151
static int export(AVFilterContext *ctx, StreamContext *sc, int input)
Definition: vf_signature.c:570
uint8_t confidence
Definition: signature.h:77
uint8_t
#define av_cold
Definition: attributes.h:82
AVOptions.
timestamp utils, mostly useful for debugging/logging purposes
CoarseSignature * curcoarsesig1
Definition: signature.h:117
static uint64_t get_block_sum(StreamContext *sc, uint64_t intpic[32][32], const Block *b)
Definition: vf_signature.c:112
struct FineSignature * first
Definition: signature.h:84
#define ELEMENT_COUNT
Definition: signature.h:36
int64_t pts
Presentation timestamp in time_base units (time when frame should be shown to user).
Definition: frame.h:294
planar YUV 4:4:0 full scale (JPEG), deprecated in favor of AV_PIX_FMT_YUV440P and setting color_range...
Definition: pixfmt.h:101
static int flags
Definition: log.c:57
planar YUV 4:2:2, 16bpp, full scale (JPEG), deprecated in favor of AV_PIX_FMT_YUV422P and setting col...
Definition: pixfmt.h:75
#define AVERROR_EOF
End of file.
Definition: error.h:55
#define av_log(a,...)
uint32_t lastindex
Definition: signature.h:122
struct FineSignature * last
Definition: signature.h:85
static int get_block_size(const Block *b)
Definition: vf_signature.c:107
A filter pad used for either input or output.
Definition: internal.h:54
#define AV_LOG_ERROR
Something went wrong and cannot losslessly be recovered.
Definition: log.h:176
int ff_set_common_formats(AVFilterContext *ctx, AVFilterFormats *formats)
A helper for query_formats() which sets all links to the same list of formats.
Definition: formats.c:568
static int config_output(AVFilterLink *outlink)
Definition: vf_signature.c:735
AVFILTER_DEFINE_CLASS(signature)
static const uint16_t mask[17]
Definition: lzw.c:38
FineSignature * curfinesig
Definition: signature.h:112
#define AVERROR(e)
Definition: error.h:43
#define NULL_IF_CONFIG_SMALL(x)
Return NULL if CONFIG_SMALL is true, otherwise the argument without modification. ...
Definition: internal.h:179
void * priv
private data for use by the filter
Definition: avfilter.h:353
#define AV_LOG_DEBUG
Stuff which is only useful for libav* developers.
Definition: log.h:197
planar YUV 4:2:0, 12bpp, 1 plane for Y and 1 plane for the UV components, which are interleaved (firs...
Definition: pixfmt.h:90
uint8_t x
Definition: signature.h:55
int av_log_get_level(void)
Get the current log level.
Definition: log.c:386
static MatchingInfo lookup_signatures(AVFilterContext *ctx, SignatureContext *sc, StreamContext *first, StreamContext *second, int mode)
static int put_bits_count(PutBitContext *s)
Definition: put_bits.h:85
planar YUV 4:2:2, 16bpp, (1 Cr & Cb sample per 2x1 Y samples)
Definition: pixfmt.h:66
char * av_asprintf(const char *fmt,...)
Definition: avstring.c:113
#define BLOCK_LCM
Definition: vf_signature.c:41
short elem_count
Definition: signature.h:68
as above, but U and V bytes are swapped
Definition: pixfmt.h:91
#define th
Definition: regdef.h:75
int matchframes
Definition: signature.h:95
struct FineSignature * second
Definition: signature.h:98
#define FFMIN(a, b)
Definition: common.h:96
planar YUV 4:2:0, 12bpp, full scale (JPEG), deprecated in favor of AV_PIX_FMT_YUV420P and setting col...
Definition: pixfmt.h:74
AVRational time_base
Definition: signature.h:103
static const ElemCat * elements[ELEMENT_COUNT]
Definition: signature.h:566
int32_t
AVFormatContext * ctx
Definition: movenc.c:48
MPEG-7 video signature calculation and lookup filter.
#define FFABS(a)
Absolute value, Note, INT_MIN / INT64_MIN result in undefined behavior as they are not representable ...
Definition: common.h:72
int n
Definition: avisynth_c.h:684
static const AVFilterPad outputs[]
Definition: af_afftfilt.c:389
short block_count
Definition: signature.h:67
int av_get_frame_filename(char *buf, int buf_size, const char *path, int number)
Definition: utils.c:4612
#define AV_LOG_INFO
Standard information.
Definition: log.h:187
static const AVFilterPad inputs[]
Definition: af_afftfilt.c:379
#define DIFFELEM_SIZE
Definition: signature.h:38
static int request_frame(AVFilterLink *outlink)
Definition: vf_signature.c:588
int linesize[AV_NUM_DATA_POINTERS]
For video, size in bytes of each picture line.
Definition: frame.h:232
short left_count
Definition: signature.h:66
static av_cold void uninit(AVFilterContext *ctx)
Definition: vf_signature.c:700
void * buf
Definition: avisynth_c.h:690
planar YUV 4:1:0, 9bpp, (1 Cr & Cb sample per 4x4 Y samples)
Definition: pixfmt.h:68
static const char * format
Definition: movenc.c:47
Filter definition.
Definition: avfilter.h:144
static int config_input(AVFilterLink *inlink)
Definition: vf_signature.c:90
const char * name
Filter name.
Definition: avfilter.h:148
AVFilterLink ** outputs
array of pointers to output links
Definition: avfilter.h:350
struct CoarseSignature * next
Definition: signature.h:86
MPEG-7 video signature calculation and lookup filter.
static enum AVPixelFormat pix_fmts[]
Definition: libkvazaar.c:266
CoarseSignature * coarsesiglist
Definition: signature.h:114
uint8_t * data[AV_NUM_DATA_POINTERS]
pointer to the picture/channel planes.
Definition: frame.h:215
int av_strerror(int errnum, char *errbuf, size_t errbuf_size)
Put a description of the AVERROR code errnum in errbuf.
Definition: error.c:105
#define FF_INLINK_IDX(link)
Find the index of a link.
Definition: internal.h:348
struct FineSignature * first
Definition: signature.h:97
static int xml_export(AVFilterContext *ctx, StreamContext *sc, const char *filename)
Definition: vf_signature.c:385
CoarseSignature * curcoarsesig2
Definition: signature.h:118
Main libavformat public API header.
int
planar YUV 4:2:0, 12bpp, (1 Cr & Cb sample per 2x2 Y samples)
Definition: pixfmt.h:62
Y , 8bpp.
Definition: pixfmt.h:70
static int query_formats(AVFilterContext *ctx)
Definition: vf_signature.c:72
static void flush_put_bits(PutBitContext *s)
Pad the end of the output stream with zeros.
Definition: put_bits.h:101
#define SIGELEM_SIZE
Definition: signature.h:37
AVFilter ff_vf_signature
Definition: vf_signature.c:759
uint64_t pts
Definition: signature.h:75
struct FineSignature * prev
Definition: signature.h:74
planar YUV 4:4:4, 24bpp, full scale (JPEG), deprecated in favor of AV_PIX_FMT_YUV444P and setting col...
Definition: pixfmt.h:76
planar YUV 4:1:1, 12bpp, (1 Cr & Cb sample per 4x1 Y samples)
Definition: pixfmt.h:69
static void init_put_bits(PutBitContext *s, uint8_t *buffer, int buffer_size)
Initialize the PutBitContext s.
Definition: put_bits.h:48
int den
Denominator.
Definition: rational.h:60
CoarseSignature * coarseend
Definition: signature.h:115
uint8_t data[5][31]
Definition: signature.h:83
int len
static av_cold int init(AVFilterContext *ctx)
Definition: vf_signature.c:644
planar YUV 4:1:1, 12bpp, (1 Cr & Cb sample per 4x1 Y samples) full scale (JPEG), deprecated in favor ...
Definition: pixfmt.h:272
An instance of a filter.
Definition: avfilter.h:338
#define FLAGS
Definition: vf_signature.c:40
Point to
Definition: signature.h:61
#define av_freep(p)
planar YUV 4:4:0 (1 Cr & Cb sample per 1x2 Y samples)
Definition: pixfmt.h:100
uint8_t words[5]
Definition: signature.h:78
#define av_malloc_array(a, b)
int lookup
int ff_request_frame(AVFilterLink *link)
Request an input frame from the filter at the other end of the link.
Definition: avfilter.c:405
internal API functions
AVPixelFormat
Pixel format.
Definition: pixfmt.h:60
Point up
Definition: signature.h:60
mode
Use these values in ebur128_init (or'ed).
Definition: ebur128.h:83
for(j=16;j >0;--j)
GLuint buffer
Definition: opengl_enc.c:102
uint8_t framesig[SIGELEM_SIZE/5]
Definition: signature.h:79
static int binary_export(AVFilterContext *ctx, StreamContext *sc, const char *filename)
Definition: vf_signature.c:491
FineSignature * finesiglist
Definition: signature.h:111
static int ff_insert_inpad(AVFilterContext *f, unsigned index, AVFilterPad *p)
Insert a new input pad for the filter.
Definition: internal.h:277
static uint8_t tmp[11]
Definition: aes_ctr.c:26
bitstream writer API