1 /*
2 * Copyright (c) 2002 Jindrich Makovicka <makovick@gmail.com>
3 * Copyright (c) 2011 Stefano Sabatini
4 * Copyright (c) 2013, 2015 Jean Delvare <jdelvare@suse.com>
5 *
6 * This file is part of FFmpeg.
7 *
8 * FFmpeg is free software; you can redistribute it and/or modify
9 * it under the terms of the GNU General Public License as published by
10 * the Free Software Foundation; either version 2 of the License, or
11 * (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
16 * GNU General Public License for more details.
17 *
18 * You should have received a copy of the GNU General Public License along
19 * with FFmpeg; if not, write to the Free Software Foundation, Inc.,
20 * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
21 */
22
23 /**
24 * @file
25 * A very simple tv station logo remover
26 * Originally imported from MPlayer libmpcodecs/vf_delogo.c,
27 * the algorithm was later improved.
28 */
29
30 #include "libavutil/common.h"
31 #include "libavutil/imgutils.h"
32 #include "libavutil/opt.h"
33 #include "libavutil/pixdesc.h"
34 #include "libavutil/eval.h"
35 #include "avfilter.h"
36 #include "formats.h"
37 #include "internal.h"
38 #include "video.h"
39 static const char * const var_names[] = {
40 "x",
41 "y",
42 "w",
43 "h",
44 "n", ///< number of frame
45 "t", ///< timestamp expressed in seconds
46 NULL
47 };
48
49 enum var_name {
50 VAR_X,
51 VAR_Y,
52 VAR_W,
53 VAR_H,
54 VAR_N,
55 VAR_T,
56 VAR_VARS_NB
57 };
58
set_expr(AVExpr ** pexpr,const char * expr,const char * option,void * log_ctx)59 static int set_expr(AVExpr **pexpr, const char *expr, const char *option, void *log_ctx)
60 {
61 int ret;
62 AVExpr *old = NULL;
63
64 if (*pexpr)
65 old = *pexpr;
66 ret = av_expr_parse(pexpr, expr, var_names, NULL, NULL, NULL, NULL, 0, log_ctx);
67 if (ret < 0) {
68 av_log(log_ctx, AV_LOG_ERROR,
69 "Error when parsing the expression '%s' for %s\n",
70 expr, option);
71 *pexpr = old;
72 return ret;
73 }
74
75 av_expr_free(old);
76 return 0;
77 }
78
79
80 /**
81 * Apply a simple delogo algorithm to the image in src and put the
82 * result in dst.
83 *
84 * The algorithm is only applied to the region specified by the logo
85 * parameters.
86 *
87 * @param w width of the input image
88 * @param h height of the input image
89 * @param logo_x x coordinate of the top left corner of the logo region
90 * @param logo_y y coordinate of the top left corner of the logo region
91 * @param logo_w width of the logo
92 * @param logo_h height of the logo
93 * @param band the size of the band around the processed area
94 * @param show show a rectangle around the processed area, useful for
95 * parameters tweaking
96 * @param direct if non-zero perform in-place processing
97 */
apply_delogo(uint8_t * dst,int dst_linesize,uint8_t * src,int src_linesize,int w,int h,AVRational sar,int logo_x,int logo_y,int logo_w,int logo_h,unsigned int band,int show,int direct)98 static void apply_delogo(uint8_t *dst, int dst_linesize,
99 uint8_t *src, int src_linesize,
100 int w, int h, AVRational sar,
101 int logo_x, int logo_y, int logo_w, int logo_h,
102 unsigned int band, int show, int direct)
103 {
104 int x, y;
105 uint64_t interp, weightl, weightr, weightt, weightb, weight;
106 uint8_t *xdst, *xsrc;
107
108 uint8_t *topleft, *botleft, *topright;
109 unsigned int left_sample, right_sample;
110 int xclipl, xclipr, yclipt, yclipb;
111 int logo_x1, logo_x2, logo_y1, logo_y2;
112
113 xclipl = FFMAX(-logo_x, 0);
114 xclipr = FFMAX(logo_x+logo_w-w, 0);
115 yclipt = FFMAX(-logo_y, 0);
116 yclipb = FFMAX(logo_y+logo_h-h, 0);
117
118 logo_x1 = logo_x + xclipl;
119 logo_x2 = logo_x + logo_w - xclipr - 1;
120 logo_y1 = logo_y + yclipt;
121 logo_y2 = logo_y + logo_h - yclipb - 1;
122
123 topleft = src+logo_y1 * src_linesize+logo_x1;
124 topright = src+logo_y1 * src_linesize+logo_x2;
125 botleft = src+logo_y2 * src_linesize+logo_x1;
126
127 if (!direct)
128 av_image_copy_plane(dst, dst_linesize, src, src_linesize, w, h);
129
130 dst += (logo_y1 + 1) * dst_linesize;
131 src += (logo_y1 + 1) * src_linesize;
132
133 for (y = logo_y1+1; y < logo_y2; y++) {
134 left_sample = topleft[src_linesize*(y-logo_y1)] +
135 topleft[src_linesize*(y-logo_y1-1)] +
136 topleft[src_linesize*(y-logo_y1+1)];
137 right_sample = topright[src_linesize*(y-logo_y1)] +
138 topright[src_linesize*(y-logo_y1-1)] +
139 topright[src_linesize*(y-logo_y1+1)];
140
141 for (x = logo_x1+1,
142 xdst = dst+logo_x1+1,
143 xsrc = src+logo_x1+1; x < logo_x2; x++, xdst++, xsrc++) {
144
145 if (show && (y == logo_y1+1 || y == logo_y2-1 ||
146 x == logo_x1+1 || x == logo_x2-1)) {
147 *xdst = 0;
148 continue;
149 }
150
151 /* Weighted interpolation based on relative distances, taking SAR into account */
152 weightl = (uint64_t) (logo_x2-x) * (y-logo_y1) * (logo_y2-y) * sar.den;
153 weightr = (uint64_t)(x-logo_x1) * (y-logo_y1) * (logo_y2-y) * sar.den;
154 weightt = (uint64_t)(x-logo_x1) * (logo_x2-x) * (logo_y2-y) * sar.num;
155 weightb = (uint64_t)(x-logo_x1) * (logo_x2-x) * (y-logo_y1) * sar.num;
156
157 interp =
158 left_sample * weightl
159 +
160 right_sample * weightr
161 +
162 (topleft[x-logo_x1] +
163 topleft[x-logo_x1-1] +
164 topleft[x-logo_x1+1]) * weightt
165 +
166 (botleft[x-logo_x1] +
167 botleft[x-logo_x1-1] +
168 botleft[x-logo_x1+1]) * weightb;
169 weight = (weightl + weightr + weightt + weightb) * 3U;
170 interp = (interp + (weight >> 1)) / weight;
171
172 if (y >= logo_y+band && y < logo_y+logo_h-band &&
173 x >= logo_x+band && x < logo_x+logo_w-band) {
174 *xdst = interp;
175 } else {
176 unsigned dist = 0;
177
178 if (x < logo_x+band)
179 dist = FFMAX(dist, logo_x-x+band);
180 else if (x >= logo_x+logo_w-band)
181 dist = FFMAX(dist, x-(logo_x+logo_w-1-band));
182
183 if (y < logo_y+band)
184 dist = FFMAX(dist, logo_y-y+band);
185 else if (y >= logo_y+logo_h-band)
186 dist = FFMAX(dist, y-(logo_y+logo_h-1-band));
187
188 *xdst = (*xsrc*dist + interp*(band-dist))/band;
189 }
190 }
191
192 dst += dst_linesize;
193 src += src_linesize;
194 }
195 }
196
197 typedef struct DelogoContext {
198 const AVClass *class;
199 int x, y, w, h, band, show;
200 char *x_expr, *y_expr, *w_expr, *h_expr;
201 AVExpr *x_pexpr, *y_pexpr, *w_pexpr, *h_pexpr;
202 double var_values[VAR_VARS_NB];
203 } DelogoContext;
204
205 #define OFFSET(x) offsetof(DelogoContext, x)
206 #define FLAGS AV_OPT_FLAG_FILTERING_PARAM|AV_OPT_FLAG_VIDEO_PARAM
207
208 static const AVOption delogo_options[]= {
209 { "x", "set logo x position", OFFSET(x_expr), AV_OPT_TYPE_STRING, { .str = "-1" }, 0, 0, FLAGS },
210 { "y", "set logo y position", OFFSET(y_expr), AV_OPT_TYPE_STRING, { .str = "-1" }, 0, 0, FLAGS },
211 { "w", "set logo width", OFFSET(w_expr), AV_OPT_TYPE_STRING, { .str = "-1" }, 0, 0, FLAGS },
212 { "h", "set logo height", OFFSET(h_expr), AV_OPT_TYPE_STRING, { .str = "-1" }, 0, 0, FLAGS },
213 { "show", "show delogo area", OFFSET(show), AV_OPT_TYPE_BOOL, { .i64 = 0 }, 0, 1, FLAGS },
214 { NULL }
215 };
216
217 AVFILTER_DEFINE_CLASS(delogo);
uninit(AVFilterContext * ctx)218 static av_cold void uninit(AVFilterContext *ctx)
219 {
220 DelogoContext *s = ctx->priv;
221
222 av_expr_free(s->x_pexpr); s->x_pexpr = NULL;
223 av_expr_free(s->y_pexpr); s->y_pexpr = NULL;
224 av_expr_free(s->w_pexpr); s->w_pexpr = NULL;
225 av_expr_free(s->h_pexpr); s->h_pexpr = NULL;
226 }
227
228 static const enum AVPixelFormat pix_fmts[] = {
229 AV_PIX_FMT_YUV444P, AV_PIX_FMT_YUV422P, AV_PIX_FMT_YUV420P,
230 AV_PIX_FMT_YUV411P, AV_PIX_FMT_YUV410P, AV_PIX_FMT_YUV440P,
231 AV_PIX_FMT_YUVA420P, AV_PIX_FMT_GRAY8,
232 AV_PIX_FMT_NONE
233 };
234
init(AVFilterContext * ctx)235 static av_cold int init(AVFilterContext *ctx)
236 {
237 DelogoContext *s = ctx->priv;
238 int ret = 0;
239
240 if ((ret = set_expr(&s->x_pexpr, s->x_expr, "x", ctx)) < 0 ||
241 (ret = set_expr(&s->y_pexpr, s->y_expr, "y", ctx)) < 0 ||
242 (ret = set_expr(&s->w_pexpr, s->w_expr, "w", ctx)) < 0 ||
243 (ret = set_expr(&s->h_pexpr, s->h_expr, "h", ctx)) < 0 )
244 return ret;
245
246 s->x = av_expr_eval(s->x_pexpr, s->var_values, s);
247 s->y = av_expr_eval(s->y_pexpr, s->var_values, s);
248 s->w = av_expr_eval(s->w_pexpr, s->var_values, s);
249 s->h = av_expr_eval(s->h_pexpr, s->var_values, s);
250
251 #define CHECK_UNSET_OPT(opt) \
252 if (s->opt == -1) { \
253 av_log(s, AV_LOG_ERROR, "Option %s was not set.\n", #opt); \
254 return AVERROR(EINVAL); \
255 }
256 CHECK_UNSET_OPT(x);
257 CHECK_UNSET_OPT(y);
258 CHECK_UNSET_OPT(w);
259 CHECK_UNSET_OPT(h);
260
261 s->band = 1;
262
263 av_log(ctx, AV_LOG_VERBOSE, "x:%d y:%d, w:%d h:%d band:%d show:%d\n",
264 s->x, s->y, s->w, s->h, s->band, s->show);
265
266 s->w += s->band*2;
267 s->h += s->band*2;
268 s->x -= s->band;
269 s->y -= s->band;
270
271 return 0;
272 }
273
config_input(AVFilterLink * inlink)274 static int config_input(AVFilterLink *inlink)
275 {
276 DelogoContext *s = inlink->dst->priv;
277
278 /* Check whether the logo area fits in the frame */
279 if (s->x + (s->band - 1) < 0 || s->x + s->w - (s->band*2 - 2) > inlink->w ||
280 s->y + (s->band - 1) < 0 || s->y + s->h - (s->band*2 - 2) > inlink->h) {
281 av_log(s, AV_LOG_ERROR, "Logo area is outside of the frame.\n");
282 return AVERROR(EINVAL);
283 }
284
285 return 0;
286 }
287
filter_frame(AVFilterLink * inlink,AVFrame * in)288 static int filter_frame(AVFilterLink *inlink, AVFrame *in)
289 {
290 DelogoContext *s = inlink->dst->priv;
291 AVFilterLink *outlink = inlink->dst->outputs[0];
292 const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(inlink->format);
293 AVFrame *out;
294 int hsub0 = desc->log2_chroma_w;
295 int vsub0 = desc->log2_chroma_h;
296 int direct = 0;
297 int plane;
298 AVRational sar;
299 int ret;
300
301 s->var_values[VAR_N] = inlink->frame_count_out;
302 s->var_values[VAR_T] = TS2T(in->pts, inlink->time_base);
303 s->x = av_expr_eval(s->x_pexpr, s->var_values, s);
304 s->y = av_expr_eval(s->y_pexpr, s->var_values, s);
305 s->w = av_expr_eval(s->w_pexpr, s->var_values, s);
306 s->h = av_expr_eval(s->h_pexpr, s->var_values, s);
307
308 if (s->x + (s->band - 1) <= 0 || s->x + s->w - (s->band*2 - 2) > inlink->w ||
309 s->y + (s->band - 1) <= 0 || s->y + s->h - (s->band*2 - 2) > inlink->h) {
310 av_log(s, AV_LOG_WARNING, "Logo area is outside of the frame,"
311 " auto set the area inside of the frame\n");
312 }
313
314 if (s->x + (s->band - 1) <= 0)
315 s->x = 1 + s->band;
316 if (s->y + (s->band - 1) <= 0)
317 s->y = 1 + s->band;
318 if (s->x + s->w - (s->band*2 - 2) > inlink->w)
319 s->w = inlink->w - s->x - (s->band*2 - 2);
320 if (s->y + s->h - (s->band*2 - 2) > inlink->h)
321 s->h = inlink->h - s->y - (s->band*2 - 2);
322
323 ret = config_input(inlink);
324 if (ret < 0) {
325 av_frame_free(&in);
326 return ret;
327 }
328
329 s->w += s->band*2;
330 s->h += s->band*2;
331 s->x -= s->band;
332 s->y -= s->band;
333
334 if (av_frame_is_writable(in)) {
335 direct = 1;
336 out = in;
337 } else {
338 out = ff_get_video_buffer(outlink, outlink->w, outlink->h);
339 if (!out) {
340 av_frame_free(&in);
341 return AVERROR(ENOMEM);
342 }
343
344 av_frame_copy_props(out, in);
345 }
346
347 sar = in->sample_aspect_ratio;
348 /* Assume square pixels if SAR is unknown */
349 if (!sar.num)
350 sar.num = sar.den = 1;
351
352 for (plane = 0; plane < desc->nb_components; plane++) {
353 int hsub = plane == 1 || plane == 2 ? hsub0 : 0;
354 int vsub = plane == 1 || plane == 2 ? vsub0 : 0;
355
356 apply_delogo(out->data[plane], out->linesize[plane],
357 in ->data[plane], in ->linesize[plane],
358 AV_CEIL_RSHIFT(inlink->w, hsub),
359 AV_CEIL_RSHIFT(inlink->h, vsub),
360 sar, s->x>>hsub, s->y>>vsub,
361 /* Up and left borders were rounded down, inject lost bits
362 * into width and height to avoid error accumulation */
363 AV_CEIL_RSHIFT(s->w + (s->x & ((1<<hsub)-1)), hsub),
364 AV_CEIL_RSHIFT(s->h + (s->y & ((1<<vsub)-1)), vsub),
365 s->band>>FFMIN(hsub, vsub),
366 s->show, direct);
367 }
368
369 if (!direct)
370 av_frame_free(&in);
371
372 return ff_filter_frame(outlink, out);
373 }
374
375 static const AVFilterPad avfilter_vf_delogo_inputs[] = {
376 {
377 .name = "default",
378 .type = AVMEDIA_TYPE_VIDEO,
379 .filter_frame = filter_frame,
380 .config_props = config_input,
381 },
382 };
383
384 static const AVFilterPad avfilter_vf_delogo_outputs[] = {
385 {
386 .name = "default",
387 .type = AVMEDIA_TYPE_VIDEO,
388 },
389 };
390
391 const AVFilter ff_vf_delogo = {
392 .name = "delogo",
393 .description = NULL_IF_CONFIG_SMALL("Remove logo from input video."),
394 .priv_size = sizeof(DelogoContext),
395 .priv_class = &delogo_class,
396 .init = init,
397 .uninit = uninit,
398 FILTER_INPUTS(avfilter_vf_delogo_inputs),
399 FILTER_OUTPUTS(avfilter_vf_delogo_outputs),
400 FILTER_PIXFMTS_ARRAY(pix_fmts),
401 .flags = AVFILTER_FLAG_SUPPORT_TIMELINE_GENERIC,
402 };
403