1. ffmpeg
  2. Commit ddcf4b5d6d19fcbf4cb632a7d9a9be4ac9e6738f
Browse code

Codebook generator using the ELBG algorithm patch by Vitor: printf(vitor%d gmail com, 1001) original thread: Re: [FFmpeg-devel] [PATCH] Add a codebook generator (was: [PATCH] RoQ video encoder, take 2) date: 05/28/2007 01:21 PM

Originally committed as revision 9196 to svn://svn.ffmpeg.org/ffmpeg/trunk

Vitor Sessak authored on 2007/06/04 16:28:34
Showing 3 changed files
Changelog
History View file @ ddcf4b5
... ... 
@@ -86,6 +86,7 @@ version <next>
86 86 
 - Renderware TXD demuxer and decoder
87 87 
 - extern C declarations for C++ removed from headers
88 88 
 - sws_flags command line option
89 
+- codebook generator
89 90
90 91 
 version 0.4.9-pre1:
91 92
libavcodec/elbg.c
History View file @ ddcf4b5
92 93 
new file mode 100644
... ... 
@@ -0,0 +1,416 @@
0 
+/*
1 
+ * Copyright (C) 2007 Vitor <vitor1001@gmail.com>
2 
+ *
3 
+ * This file is part of FFmpeg.
4 
+ *
5 
+ * FFmpeg is free software; you can redistribute it and/or
6 
+ * modify it under the terms of the GNU Lesser General Public
7 
+ * License as published by the Free Software Foundation; either
8 
+ * version 2.1 of the License, or (at your option) any later version.
9 
+ *
10 
+ * FFmpeg is distributed in the hope that it will be useful,
11 
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
12 
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
13 
+ * Lesser General Public License for more details.
14 
+ *
15 
+ * You should have received a copy of the GNU Lesser General Public
16 
+ * License along with FFmpeg; if not, write to the Free Software
17 
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
18 
+ */
19 
+
20 
+/**
21 
+ * @file cbook_gen.c
22 
+ * Codebook Generator using the ELBG algorithm
23 
+ */
24 
+
25 
+#include <string.h>
26 
+
27 
+#include "elbg.h"
28 
+#include "avcodec.h"
29 
+#include "random.h"
30 
+
31 
+#define DELTA_ERR_MAX 0.1  ///< Precision of the ELBG algorithm (as percentual error)
32 
+
33 
+/**
34 
+ * In the ELBG jargon, a cell is the set of points that are closest to a
35 
+ * codebook entry. Not to be confused with a RoQ Video cell. */
36 
+typedef struct cell_s {
37 
+    int index;
38 
+    struct cell_s *next;
39 
+} cell;
40 
+
41 
+/**
42 
+ * ELBG internal data
43 
+ */
44 
+typedef struct{
45 
+    int error;
46 
+    int dim;
47 
+    int numCB;
48 
+    int *codebook;
49 
+    cell **cells;
50 
+    int *utility;
51 
+    int *utility_inc;
52 
+    int *nearest_cb;
53 
+    int *points;
54 
+    AVRandomState *rand_state;
55 
+} elbg_data;
56 
+
57 
+static inline int distance_limited(int *a, int *b, int dim, int limit)
58 
+{
59 
+    int i, dist=0;
60 
+    for (i=0; i<dim; i++) {
61 
+        dist += (a[i] - b[i])*(a[i] - b[i]);
62 
+        if (dist > limit)
63 
+            return INT_MAX;
64 
+    }
65 
+
66 
+    return dist;
67 
+}
68 
+
69 
+static inline void vect_division(int *res, int *vect, int div, int dim)
70 
+{
71 
+    int i;
72 
+    if (div > 1)
73 
+        for (i=0; i<dim; i++)
74 
+            res[i] = ROUNDED_DIV(vect[i],div);
75 
+    else if (res != vect)
76 
+        memcpy(res, vect, dim*sizeof(int));
77 
+
78 
+}
79 
+
80 
+static int eval_error_cell(elbg_data *elbg, int *centroid, cell *cells)
81 
+{
82 
+    int error=0;
83 
+    for (; cells; cells=cells->next)
84 
+        error += distance_limited(centroid, elbg->points + cells->index*elbg->dim, elbg->dim, INT_MAX);
85 
+
86 
+    return error;
87 
+}
88 
+
89 
+static int get_closest_codebook(elbg_data *elbg, int index)
90 
+{
91 
+    int i, pick=0, diff, diff_min = INT_MAX;
92 
+    for (i=0; i<elbg->numCB; i++)
93 
+        if (i != index) {
94 
+            diff = distance_limited(elbg->codebook + i*elbg->dim, elbg->codebook + index*elbg->dim, elbg->dim, diff_min);
95 
+            if (diff < diff_min) {
96 
+                pick = i;
97 
+                diff_min = diff;
98 
+            }
99 
+        }
100 
+    return pick;
101 
+}
102 
+
103 
+static int get_high_utility_cell(elbg_data *elbg)
104 
+{
105 
+    int i=0;
106 
+    /* Using linear search, do binary if it ever turns to be speed critical */
107 
+    int r = av_random(elbg->rand_state)%elbg->utility_inc[elbg->numCB-1];
108 
+    while (elbg->utility_inc[i] < r)
109 
+        i++;
110 
+    return i;
111 
+}
112 
+
113 
+/**
114 
+ * Implementation of the simple LBG algorithm for just two codebooks
115 
+ */
116 
+static int simple_lbg(int dim,
117 
+                      int centroid[3][dim],
118 
+                      int newutility[3],
119 
+                      int *points,
120 
+                      cell *cells)
121 
+{
122 
+    int i, idx;
123 
+    int numpoints[2] = {0,0};
124 
+    int newcentroid[2][dim];
125 
+    cell *tempcell;
126 
+
127 
+    memset(newcentroid, 0, sizeof(newcentroid));
128 
+
129 
+    newutility[0] =
130 
+    newutility[1] = 0;
131 
+
132 
+    for (tempcell = cells; tempcell; tempcell=tempcell->next) {
133 
+        idx = distance_limited(centroid[0], points + tempcell->index*dim, dim, INT_MAX)>=
134 
+              distance_limited(centroid[1], points + tempcell->index*dim, dim, INT_MAX);
135 
+        numpoints[idx]++;
136 
+        for (i=0; i<dim; i++)
137 
+            newcentroid[idx][i] += points[tempcell->index*dim + i];
138 
+    }
139 
+
140 
+    vect_division(centroid[0], newcentroid[0], numpoints[0], dim);
141 
+    vect_division(centroid[1], newcentroid[1], numpoints[1], dim);
142 
+
143 
+    for (tempcell = cells; tempcell; tempcell=tempcell->next) {
144 
+        int dist[2] = {distance_limited(centroid[0], points + tempcell->index*dim, dim, INT_MAX),
145 
+                       distance_limited(centroid[1], points + tempcell->index*dim, dim, INT_MAX)};
146 
+        int idx = dist[0] > dist[1];
147 
+        newutility[idx] += dist[idx];
148 
+    }
149 
+
150 
+    return newutility[0] + newutility[1];
151 
+}
152 
+
153 
+static void get_new_centroids(elbg_data *elbg, int huc, int *newcentroid_i,
154 
+                              int *newcentroid_p)
155 
+{
156 
+    cell *tempcell;
157 
+    int min[elbg->dim];
158 
+    int max[elbg->dim];
159 
+    int i;
160 
+
161 
+    for (i=0; i< elbg->dim; i++) {
162 
+        min[i]=INT_MAX;
163 
+        max[i]=0;
164 
+    }
165 
+
166 
+    for (tempcell = elbg->cells[huc]; tempcell; tempcell = tempcell->next)
167 
+        for(i=0; i<elbg->dim; i++) {
168 
+            min[i]=FFMIN(min[i], elbg->points[tempcell->index*elbg->dim + i]);
169 
+            max[i]=FFMAX(max[i], elbg->points[tempcell->index*elbg->dim + i]);
170 
+        }
171 
+
172 
+    for (i=0; i<elbg->dim; i++) {
173 
+        newcentroid_i[i] = min[i] + (max[i] - min[i])/3;
174 
+        newcentroid_p[i] = min[i] + (2*(max[i] - min[i]))/3;
175 
+    }
176 
+}
177 
+
178 
+/**
179 
+ * Add the points in the low utility cell to its closest cell. Split the high
180 
+ * utility cell, putting the separed points in the (now empty) low utility
181 
+ * cell.
182 
+ *
183 
+ * @param elbg         Internal elbg data
184 
+ * @param indexes      {luc, huc, cluc}
185 
+ * @param newcentroid  A vector with the position of the new centroids
186 
+ */
187 
+static void shift_codebook(elbg_data *elbg, int *indexes,
188 
+                           int newcentroid[3][elbg->dim])
189 
+{
190 
+    cell *tempdata;
191 
+    cell **pp = &elbg->cells[indexes[2]];
192 
+
193 
+    while(*pp)
194 
+        pp= &(*pp)->next;
195 
+
196 
+    *pp = elbg->cells[indexes[0]];
197 
+
198 
+    elbg->cells[indexes[0]] = NULL;
199 
+    tempdata = elbg->cells[indexes[1]];
200 
+    elbg->cells[indexes[1]] = NULL;
201 
+
202 
+    while(tempdata) {
203 
+        cell *tempcell2 = tempdata->next;
204 
+        int idx = distance_limited(elbg->points + tempdata->index*elbg->dim,
205 
+                           newcentroid[0], elbg->dim, INT_MAX) >
206 
+                  distance_limited(elbg->points + tempdata->index*elbg->dim,
207 
+                           newcentroid[1], elbg->dim, INT_MAX);
208 
+
209 
+        tempdata->next = elbg->cells[indexes[idx]];
210 
+        elbg->cells[indexes[idx]] = tempdata;
211 
+        tempdata = tempcell2;
212 
+    }
213 
+}
214 
+
215 
+static void evaluate_utility_inc(elbg_data *elbg)
216 
+{
217 
+    int i, inc=0;
218 
+
219 
+    for (i=0; i < elbg->numCB; i++) {
220 
+        if (elbg->numCB*elbg->utility[i] > elbg->error)
221 
+            inc += elbg->utility[i];
222 
+        elbg->utility_inc[i] = inc;
223 
+    }
224 
+}
225 
+
226 
+
227 
+static void update_utility_and_n_cb(elbg_data *elbg, int idx, int newutility)
228 
+{
229 
+    cell *tempcell;
230 
+
231 
+    elbg->utility[idx] = newutility;
232 
+    for (tempcell=elbg->cells[idx]; tempcell; tempcell=tempcell->next)
233 
+        elbg->nearest_cb[tempcell->index] = idx;
234 
+}
235 
+
236 
+/**
237 
+ * Evaluate if a shift lower the error. If it does, call shift_codebooks
238 
+ * and update elbg->error, elbg->utility and elbg->nearest_cb.
239 
+ *
240 
+ * @param elbg  Internal elbg data
241 
+ * @param indexes      {luc (low utility cell, huc (high utility cell), cluc (closest cell to low utility cell)}
242 
+ */
243 
+static void try_shift_candidate(elbg_data *elbg, int idx[3])
244 
+{
245 
+    int j, k, olderror=0, newerror, cont=0;
246 
+    int newutility[3];
247 
+    int newcentroid[3][elbg->dim];
248 
+    cell *tempcell;
249 
+
250 
+    for (j=0; j<3; j++)
251 
+        olderror += elbg->utility[idx[j]];
252 
+
253 
+    memset(newcentroid[2], 0, elbg->dim*sizeof(int));
254 
+
255 
+    for (k=0; k<2; k++)
256 
+        for (tempcell=elbg->cells[idx[2*k]]; tempcell; tempcell=tempcell->next) {
257 
+            cont++;
258 
+            for (j=0; j<elbg->dim; j++)
259 
+                newcentroid[2][j] += elbg->points[tempcell->index*elbg->dim + j];
260 
+        }
261 
+
262 
+    vect_division(newcentroid[2], newcentroid[2], cont, elbg->dim);
263 
+
264 
+    get_new_centroids(elbg, idx[1], newcentroid[0], newcentroid[1]);
265 
+
266 
+    newutility[2]  = eval_error_cell(elbg, newcentroid[2], elbg->cells[idx[0]]);
267 
+    newutility[2] += eval_error_cell(elbg, newcentroid[2], elbg->cells[idx[2]]);
268 
+
269 
+    newerror = newutility[2];
270 
+
271 
+    newerror += simple_lbg(elbg->dim, newcentroid, newutility, elbg->points,
272 
+                           elbg->cells[idx[1]]);
273 
+
274 
+    if (olderror > newerror) {
275 
+        shift_codebook(elbg, idx, newcentroid);
276 
+
277 
+        elbg->error += newerror - olderror;
278 
+
279 
+        for (j=0; j<3; j++)
280 
+            update_utility_and_n_cb(elbg, idx[j], newutility[j]);
281 
+
282 
+        evaluate_utility_inc(elbg);
283 
+    }
284 
+ }
285 
+
286 
+/**
287 
+ * Implementation of the ELBG block
288 
+ */
289 
+static void do_shiftings(elbg_data *elbg)
290 
+{
291 
+    int idx[3];
292 
+
293 
+    evaluate_utility_inc(elbg);
294 
+
295 
+    for (idx[0]=0; idx[0] < elbg->numCB; idx[0]++)
296 
+        if (elbg->numCB*elbg->utility[idx[0]] < elbg->error) {
297 
+            if (elbg->utility_inc[elbg->numCB-1] == 0)
298 
+                return;
299 
+
300 
+            idx[1] = get_high_utility_cell(elbg);
301 
+            idx[2] = get_closest_codebook(elbg, idx[0]);
302 
+
303 
+            try_shift_candidate(elbg, idx);
304 
+        }
305 
+}
306 
+
307 
+#define BIG_PRIME 433494437LL
308 
+
309 
+void ff_init_elbg(int *points, int dim, int numpoints, int *codebook,
310 
+                  int numCB, int max_steps, int *closest_cb,
311 
+                  AVRandomState *rand_state)
312 
+{
313 
+    int i, k;
314 
+
315 
+    if (numpoints > 24*numCB) {
316 
+        /* ELBG is very costly for a big number of points. So if we have a lot
317 
+           of them, get a good initial codebook to save on iterations       */
318 
+        int *temp_points = av_malloc(dim*(numpoints/8)*sizeof(int));
319 
+        for (i=0; i<numpoints/8; i++) {
320 
+            k = (i*BIG_PRIME) % numpoints;
321 
+            memcpy(temp_points + i*dim, points + k*dim, dim*sizeof(int));
322 
+        }
323 
+
324 
+        ff_init_elbg(temp_points, dim, numpoints/8, codebook, numCB, 2*max_steps, closest_cb, rand_state);
325 
+        ff_do_elbg(temp_points, dim, numpoints/8, codebook, numCB, 2*max_steps, closest_cb, rand_state);
326 
+
327 
+        av_free(temp_points);
328 
+
329 
+    } else  // If not, initialize the codebook with random positions
330 
+        for (i=0; i < numCB; i++)
331 
+            memcpy(codebook + i*dim, points + ((i*BIG_PRIME)%numpoints)*dim,
332 
+                   dim*sizeof(int));
333 
+
334 
+}
335 
+
336 
+void ff_do_elbg(int *points, int dim, int numpoints, int *codebook,
337 
+                int numCB, int max_steps, int *closest_cb,
338 
+                AVRandomState *rand_state)
339 
+{
340 
+    int dist;
341 
+    elbg_data elbg_d;
342 
+    elbg_data *elbg = &elbg_d;
343 
+    int i, j, k, last_error, steps=0;
344 
+    int *dist_cb = av_malloc(numpoints*sizeof(int));
345 
+    int *size_part = av_malloc(numCB*sizeof(int));
346 
+    cell *list_buffer = av_malloc(numpoints*sizeof(cell));
347 
+    cell *free_cells;
348 
+
349 
+    elbg->error = INT_MAX;
350 
+    elbg->dim = dim;
351 
+    elbg->numCB = numCB;
352 
+    elbg->codebook = codebook;
353 
+    elbg->cells = av_malloc(numCB*sizeof(cell *));
354 
+    elbg->utility = av_malloc(numCB*sizeof(int));
355 
+    elbg->nearest_cb = closest_cb;
356 
+    elbg->points = points;
357 
+    elbg->utility_inc = av_malloc(numCB*sizeof(int));
358 
+
359 
+    elbg->rand_state = rand_state;
360 
+
361 
+    do {
362 
+        free_cells = list_buffer;
363 
+        last_error = elbg->error;
364 
+        steps++;
365 
+        memset(elbg->utility, 0, numCB*sizeof(int));
366 
+        memset(elbg->cells, 0, numCB*sizeof(cell *));
367 
+
368 
+        elbg->error = 0;
369 
+
370 
+        /* This loop evaluate the actual Voronoi partition. It is the most
371 
+           costly part of the algorithm. */
372 
+        for (i=0; i < numpoints; i++) {
373 
+            dist_cb[i] = INT_MAX;
374 
+            for (k=0; k < elbg->numCB; k++) {
375 
+                dist = distance_limited(elbg->points + i*elbg->dim, elbg->codebook + k*elbg->dim, dim, dist_cb[i]);
376 
+                if (dist < dist_cb[i]) {
377 
+                    dist_cb[i] = dist;
378 
+                    elbg->nearest_cb[i] = k;
379 
+                }
380 
+            }
381 
+            elbg->error += dist_cb[i];
382 
+            elbg->utility[elbg->nearest_cb[i]] += dist_cb[i];
383 
+            free_cells->index = i;
384 
+            free_cells->next = elbg->cells[elbg->nearest_cb[i]];
385 
+            elbg->cells[elbg->nearest_cb[i]] = free_cells;
386 
+            free_cells++;
387 
+        }
388 
+
389 
+        do_shiftings(elbg);
390 
+
391 
+        memset(size_part, 0, numCB*sizeof(int));
392 
+
393 
+        memset(elbg->codebook, 0, elbg->numCB*dim*sizeof(int));
394 
+
395 
+        for (i=0; i < numpoints; i++) {
396 
+            size_part[elbg->nearest_cb[i]]++;
397 
+            for (j=0; j < elbg->dim; j++)
398 
+                elbg->codebook[elbg->nearest_cb[i]*elbg->dim + j] +=
399 
+                    elbg->points[i*elbg->dim + j];
400 
+        }
401 
+
402 
+        for (i=0; i < elbg->numCB; i++)
403 
+            vect_division(elbg->codebook + i*elbg->dim,
404 
+                          elbg->codebook + i*elbg->dim, size_part[i], elbg->dim);
405 
+
406 
+    } while(((last_error - elbg->error) > DELTA_ERR_MAX*elbg->error) &&
407 
+            (steps < max_steps));
408 
+
409 
+    av_free(dist_cb);
410 
+    av_free(size_part);
411 
+    av_free(elbg->utility);
412 
+    av_free(list_buffer);
413 
+    av_free(elbg->cells);
414 
+    av_free(elbg->utility_inc);
415 
+}
libavcodec/elbg.h
History View file @ ddcf4b5
0 416 
new file mode 100644
... ... 
@@ -0,0 +1,50 @@
0 
+/*
1 
+ * Copyright (C) 2007 Vitor <vitor1001@gmail.com>
2 
+ *
3 
+ * This file is part of FFmpeg.
4 
+ *
5 
+ * FFmpeg is free software; you can redistribute it and/or
6 
+ * modify it under the terms of the GNU Lesser General Public
7 
+ * License as published by the Free Software Foundation; either
8 
+ * version 2.1 of the License, or (at your option) any later version.
9 
+ *
10 
+ * FFmpeg is distributed in the hope that it will be useful,
11 
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
12 
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
13 
+ * Lesser General Public License for more details.
14 
+ *
15 
+ * You should have received a copy of the GNU Lesser General Public
16 
+ * License along with FFmpeg; if not, write to the Free Software
17 
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
18 
+ */
19 
+
20 
+#include "random.h"
21 
+
22 
+/**
23 
+ * Implementation of the Enhanced LBG Algorithm
24 
+ * Based on the paper "Neural Networks 14:1219-1237" that can be found in
25 
+ * http://citeseer.ist.psu.edu/patan01enhanced.html .
26 
+ *
27 
+ * @param points Input points.
28 
+ * @param dim Dimension of the points.
29 
+ * @param numpoints Num of points in **points.
30 
+ * @param codebook Pointer to the output codebook. Must be allocated.
31 
+ * @param numCB Number of points in the codebook.
32 
+ * @param num_steps The maximum number of steps. One step is already a good compromise between time and quality.
33 
+ * @param closest_cb Return the closest codebook to each point. Must be allocated.
34 
+ * @param rand_state A random number generator state. Should be already initialised by av_init_random.
35 
+ */
36 
+void ff_do_elbg(int *points, int dim, int numpoints, int *codebook,
37 
+                int numCB, int num_steps, int *closest_cb,
38 
+                AVRandomState *rand_state);
39 
+
40 
+/**
41 
+ * Initialize the **codebook vector for the elbg algorithm. If you have already
42 
+ * a codebook and you want to refine it, you shouldn't call this function.
43 
+ * If numpoints < 8*numCB this function fills **codebook with random numbers.
44 
+ * If not, it calls ff_do_elbg for a (smaller) random sample of the points in
45 
+ * **points. Get the same parameters as ff_do_elbg.
46 
+ */
47 
+void ff_init_elbg(int *points, int dim, int numpoints, int *codebook,
48 
+                  int numCB, int num_steps, int *closest_cb,
49 
+                  AVRandomState *rand_state);