FFmpeg coverage

Directory: ../../../ffmpeg/
File: src/libavfilter/vf_find_rect.c
Date: 2026-04-19 20:43:40
Exec Total Coverage
Lines: 0 139 0.0%
Functions: 0 6 0.0%
Branches: 0 66 0.0%
Line Branch Exec Source
1 /*
2 * Copyright (c) 2014-2015 Michael Niedermayer <michaelni@gmx.at>
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 * @todo switch to dualinput
23 */
24
25 #include "libavutil/opt.h"
26
27 #include "filters.h"
28 #include "video.h"
29
30 #include "lavfutils.h"
31
32 #define MAX_MIPMAPS 5
33
34 typedef struct FOCContext {
35 AVClass *class;
36 float threshold;
37 int mipmaps;
38 int xmin, ymin, xmax, ymax;
39 char *obj_filename;
40 int last_x, last_y;
41 AVFrame *obj_frame;
42 AVFrame *needle_frame[MAX_MIPMAPS];
43 AVFrame *haystack_frame[MAX_MIPMAPS];
44 int discard;
45 } FOCContext;
46
47 #define OFFSET(x) offsetof(FOCContext, x)
48 #define FLAGS AV_OPT_FLAG_FILTERING_PARAM|AV_OPT_FLAG_VIDEO_PARAM
49 static const AVOption find_rect_options[] = {
50 { "object", "object bitmap filename", OFFSET(obj_filename), AV_OPT_TYPE_STRING, {.str = NULL}, .flags = FLAGS },
51 { "threshold", "set threshold", OFFSET(threshold), AV_OPT_TYPE_FLOAT, {.dbl = 0.5}, 0, 1.0, FLAGS },
52 { "mipmaps", "set mipmaps", OFFSET(mipmaps), AV_OPT_TYPE_INT, {.i64 = 3}, 1, MAX_MIPMAPS, FLAGS },
53 { "xmin", "", OFFSET(xmin), AV_OPT_TYPE_INT, {.i64 = 0}, 0, INT_MAX, FLAGS },
54 { "ymin", "", OFFSET(ymin), AV_OPT_TYPE_INT, {.i64 = 0}, 0, INT_MAX, FLAGS },
55 { "xmax", "", OFFSET(xmax), AV_OPT_TYPE_INT, {.i64 = INT_MAX}, 0, INT_MAX, FLAGS },
56 { "ymax", "", OFFSET(ymax), AV_OPT_TYPE_INT, {.i64 = INT_MAX}, 0, INT_MAX, FLAGS },
57 { "discard", "", OFFSET(discard), AV_OPT_TYPE_BOOL, {.i64 = 0}, 0, 1, FLAGS },
58 { NULL }
59 };
60
61 AVFILTER_DEFINE_CLASS(find_rect);
62
63 ✗ static AVFrame *downscale(AVFrame *in)
64 {
65 int x, y;
66 ✗ AVFrame *frame = av_frame_alloc();
67 uint8_t *src, *dst;
68 ✗ if (!frame)
69 ✗ return NULL;
70
71 ✗ frame->format = in->format;
72 ✗ frame->width = (in->width + 1) / 2;
73 ✗ frame->height = (in->height+ 1) / 2;
74
75 ✗ if (av_frame_get_buffer(frame, 0) < 0) {
76 ✗ av_frame_free(&frame);
77 ✗ return NULL;
78 }
79 ✗ src = in ->data[0];
80 ✗ dst = frame->data[0];
81
82 ✗ int w2 = in->width/2;
83 ✗ int h2 = in->height/2;
84 ✗ for(y = 0; y < h2; y++) {
85 ✗ for(x = 0; x < w2; x++) {
86 ✗ dst[x] = ( src[2*x+0]
87 ✗ + src[2*x+1]
88 ✗ + src[2*x+0 + in->linesize[0]]
89 ✗ + src[2*x+1 + in->linesize[0]]
90 ✗ + 2) >> 2;
91 }
92 ✗ src += 2*in->linesize[0];
93 ✗ dst += frame->linesize[0];
94 }
95 ✗ src = in ->data[0];
96 ✗ dst = frame->data[0];
97 ✗ for(y = 0; y < frame->height; y++) {
98 ✗ int yd = y < h2 ? in->linesize[0] : 0;
99 ✗ x = yd ? w2 : 0;
100 ✗ for(; x < frame->width; x++) {
101 ✗ dst[x] = ( src[2*x+0]
102 ✗ + src[FFMIN(2*x+1, w2)]
103 ✗ + src[2*x+0 + yd]
104 ✗ + src[FFMIN(2*x+1, w2) + yd]
105 ✗ + 2) >> 2;
106 }
107 ✗ src += 2*in->linesize[0];
108 ✗ dst += frame->linesize[0];
109 }
110
111 ✗ return frame;
112 }
113
114 ✗ static float compare(const AVFrame *haystack, const AVFrame *obj, int offx, int offy)
115 {
116 int x,y;
117 ✗ int o_sum_v = 0;
118 ✗ int h_sum_v = 0;
119 ✗ int64_t oo_sum_v = 0;
120 ✗ int64_t hh_sum_v = 0;
121 ✗ int64_t oh_sum_v = 0;
122 float c;
123 ✗ int n = obj->height * obj->width;
124 ✗ const uint8_t *odat = obj ->data[0];
125 ✗ const uint8_t *hdat = haystack->data[0] + offx + offy * haystack->linesize[0];
126 int64_t o_sigma, h_sigma;
127
128 ✗ for(y = 0; y < obj->height; y++) {
129 ✗ for(x = 0; x < obj->width; x++) {
130 ✗ int o_v = odat[x];
131 ✗ int h_v = hdat[x];
132 ✗ o_sum_v += o_v;
133 ✗ h_sum_v += h_v;
134 ✗ oo_sum_v += o_v * o_v;
135 ✗ hh_sum_v += h_v * h_v;
136 ✗ oh_sum_v += o_v * h_v;
137 }
138 ✗ odat += obj->linesize[0];
139 ✗ hdat += haystack->linesize[0];
140 }
141 ✗ o_sigma = n*oo_sum_v - o_sum_v*(int64_t)o_sum_v;
142 ✗ h_sigma = n*hh_sum_v - h_sum_v*(int64_t)h_sum_v;
143
144 ✗ if (o_sigma == 0 || h_sigma == 0)
145 ✗ return 1.0;
146
147 ✗ c = (n*oh_sum_v - o_sum_v*(int64_t)h_sum_v) / (sqrt(o_sigma)*sqrt(h_sigma));
148
149 ✗ return 1 - fabs(c);
150 }
151
152 ✗ static float search(FOCContext *foc, int pass, int maxpass, int xmin, int xmax, int ymin, int ymax, int *best_x, int *best_y, float best_score)
153 {
154 int x, y;
155
156 ✗ if (pass + 1 <= maxpass) {
157 int sub_x, sub_y;
158 ✗ search(foc, pass+1, maxpass, xmin>>1, (xmax+1)>>1, ymin>>1, (ymax+1)>>1, &sub_x, &sub_y, 2.0);
159 ✗ xmin = FFMAX(xmin, 2*sub_x - 4);
160 ✗ xmax = FFMIN(xmax, 2*sub_x + 4);
161 ✗ ymin = FFMAX(ymin, 2*sub_y - 4);
162 ✗ ymax = FFMIN(ymax, 2*sub_y + 4);
163 }
164
165 ✗ for (y = ymin; y <= ymax; y++) {
166 ✗ for (x = xmin; x <= xmax; x++) {
167 ✗ float score = compare(foc->haystack_frame[pass], foc->needle_frame[pass], x, y);
168 ✗ if (score < best_score) {
169 ✗ best_score = score;
170 ✗ *best_x = x;
171 ✗ *best_y = y;
172 }
173 }
174 }
175 ✗ return best_score;
176 }
177
178 ✗ static int filter_frame(AVFilterLink *inlink, AVFrame *in)
179 {
180 ✗ FilterLink *inl = ff_filter_link(inlink);
181 ✗ AVFilterContext *ctx = inlink->dst;
182 ✗ FOCContext *foc = ctx->priv;
183 float best_score;
184 int best_x, best_y;
185 int i;
186 char buf[32];
187
188 ✗ int xmin = FFMAX(foc->xmin, 0);
189 ✗ int ymin = FFMAX(foc->ymin, 0);
190 ✗ int xmax = FFMIN(foc->xmax, inlink->w - foc->obj_frame->width );
191 ✗ int ymax = FFMIN(foc->ymax, inlink->h - foc->obj_frame->height);
192
193 ✗ if (xmin > xmax || ymin > ymax)
194 ✗ av_log(ctx, AV_LOG_WARNING, "x/y min/max are invalid for the current frame\n");
195
196
197 ✗ foc->haystack_frame[0] = av_frame_clone(in);
198 ✗ for (i=1; i<foc->mipmaps; i++) {
199 ✗ foc->haystack_frame[i] = downscale(foc->haystack_frame[i-1]);
200 }
201
202 ✗ best_score = search(foc, 0, 0,
203 ✗ FFMAX(xmin, foc->last_x - 8),
204 ✗ FFMIN(xmax, foc->last_x + 8),
205 ✗ FFMAX(ymin, foc->last_y - 8),
206 ✗ FFMIN(ymax, foc->last_y + 8),
207 &best_x, &best_y, 2.0);
208
209 ✗ best_score = search(foc, 0, foc->mipmaps - 1, xmin, xmax, ymin, ymax,
210 &best_x, &best_y, best_score);
211
212 ✗ for (i=0; i<MAX_MIPMAPS; i++) {
213 ✗ av_frame_free(&foc->haystack_frame[i]);
214 }
215
216 ✗ if (best_score > foc->threshold) {
217 ✗ if (foc->discard) {
218 ✗ av_frame_free(&in);
219 ✗ return 0;
220 } else {
221 ✗ return ff_filter_frame(ctx->outputs[0], in);
222 }
223 }
224
225 ✗ av_log(ctx, AV_LOG_INFO, "Found at n=%"PRId64" pts_time=%f x=%d y=%d with score=%f\n",
226 ✗ inl->frame_count_out, TS2D(in->pts) * av_q2d(inlink->time_base),
227 best_x, best_y, best_score);
228 ✗ foc->last_x = best_x;
229 ✗ foc->last_y = best_y;
230
231 ✗ snprintf(buf, sizeof(buf), "%f", best_score);
232
233 ✗ av_dict_set_int(&in->metadata, "lavfi.rect.w", foc->obj_frame->width, 0);
234 ✗ av_dict_set_int(&in->metadata, "lavfi.rect.h", foc->obj_frame->height, 0);
235 ✗ av_dict_set_int(&in->metadata, "lavfi.rect.x", best_x, 0);
236 ✗ av_dict_set_int(&in->metadata, "lavfi.rect.y", best_y, 0);
237 ✗ av_dict_set(&in->metadata, "lavfi.rect.score", buf, 0);
238
239 ✗ return ff_filter_frame(ctx->outputs[0], in);
240 }
241
242 ✗ static av_cold void uninit(AVFilterContext *ctx)
243 {
244 ✗ FOCContext *foc = ctx->priv;
245 int i;
246
247 ✗ for (i = 0; i < MAX_MIPMAPS; i++) {
248 ✗ av_frame_free(&foc->needle_frame[i]);
249 ✗ av_frame_free(&foc->haystack_frame[i]);
250 }
251
252 ✗ av_frame_free(&foc->obj_frame);
253 ✗ }
254
255 ✗ static av_cold int init(AVFilterContext *ctx)
256 {
257 ✗ FOCContext *foc = ctx->priv;
258 int ret, i;
259
260 ✗ if (!foc->obj_filename) {
261 ✗ av_log(ctx, AV_LOG_ERROR, "object filename not set\n");
262 ✗ return AVERROR(EINVAL);
263 }
264
265 ✗ ret = ff_load_image(&foc->obj_frame, foc->obj_filename, ctx);
266 ✗ if (ret < 0)
267 ✗ return ret;
268
269 ✗ if (foc->obj_frame->format != AV_PIX_FMT_GRAY8) {
270 ✗ av_log(ctx, AV_LOG_ERROR, "object image is not a grayscale image\n");
271 ✗ return AVERROR(EINVAL);
272 }
273
274 ✗ foc->needle_frame[0] = av_frame_clone(foc->obj_frame);
275 ✗ for (i = 1; i < foc->mipmaps; i++) {
276 ✗ foc->needle_frame[i] = downscale(foc->needle_frame[i-1]);
277 ✗ if (!foc->needle_frame[i])
278 ✗ return AVERROR(ENOMEM);
279 }
280
281 ✗ return 0;
282 }
283
284 static const AVFilterPad foc_inputs[] = {
285 {
286 .name = "default",
287 .type = AVMEDIA_TYPE_VIDEO,
288 .filter_frame = filter_frame,
289 },
290 };
291
292 const FFFilter ff_vf_find_rect = {
293 .p.name = "find_rect",
294 .p.description = NULL_IF_CONFIG_SMALL("Find a user specified object."),
295 .p.flags = AVFILTER_FLAG_METADATA_ONLY,
296 .p.priv_class = &find_rect_class,
297 .priv_size = sizeof(FOCContext),
298 .init = init,
299 .uninit = uninit,
300 FILTER_INPUTS(foc_inputs),
301 FILTER_OUTPUTS(ff_video_default_filterpad),
302 FILTER_PIXFMTS(AV_PIX_FMT_YUV420P, AV_PIX_FMT_YUVJ420P),
303 };
304