libopenjpegdec.c 14.1 KB
Newer Older
1 2 3 4
/*
 * JPEG 2000 decoding support via OpenJPEG
 * Copyright (c) 2009 Jaikrishnan Menon <realityman@gmx.net>
 *
5
 * This file is part of Libav.
6
 *
7
 * Libav is free software; you can redistribute it and/or
8 9 10 11
 * modify it under the terms of the GNU Lesser General Public
 * License as published by the Free Software Foundation; either
 * version 2.1 of the License, or (at your option) any later version.
 *
12
 * Libav is distributed in the hope that it will be useful,
13 14 15 16 17
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * Lesser General Public License for more details.
 *
 * You should have received a copy of the GNU Lesser General Public
18
 * License along with Libav; if not, write to the Free Software
19 20 21 22
 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
 */

/**
23 24 25 26 27 28
 * @file
 * JPEG 2000 decoder using libopenjpeg
 */

#define  OPJ_STATIC
#include <openjpeg.h>
29

30
#include "libavutil/common.h"
31
#include "libavutil/intreadwrite.h"
32 33
#include "libavutil/imgutils.h"
#include "libavutil/pixfmt.h"
34 35
#include "libavutil/opt.h"
#include "avcodec.h"
36
#include "thread.h"
37 38 39 40

#define JP2_SIG_TYPE    0x6A502020
#define JP2_SIG_VALUE   0x0D0A870A

41 42
// pix_fmts with lower bpp have to be listed before
// similar pix_fmts with higher bpp.
43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63
#define RGB_PIXEL_FORMATS  AV_PIX_FMT_RGB24, AV_PIX_FMT_RGBA,  \
                           AV_PIX_FMT_RGB48

#define GRAY_PIXEL_FORMATS AV_PIX_FMT_GRAY8, AV_PIX_FMT_Y400A, \
                           AV_PIX_FMT_GRAY16

#define YUV_PIXEL_FORMATS  AV_PIX_FMT_YUV410P,   AV_PIX_FMT_YUV411P,   \
                           AV_PIX_FMT_YUVA420P, \
                           AV_PIX_FMT_YUV420P,   AV_PIX_FMT_YUV422P,   \
                           AV_PIX_FMT_YUV440P,   AV_PIX_FMT_YUV444P,   \
                           AV_PIX_FMT_YUV420P9,  AV_PIX_FMT_YUV422P9,  \
                           AV_PIX_FMT_YUV444P9, \
                           AV_PIX_FMT_YUV420P10, AV_PIX_FMT_YUV422P10, \
                           AV_PIX_FMT_YUV444P10, \
                           AV_PIX_FMT_YUV420P16, AV_PIX_FMT_YUV422P16, \
                           AV_PIX_FMT_YUV444P16

static const enum AVPixelFormat rgb_pix_fmts[]  = {RGB_PIXEL_FORMATS};
static const enum AVPixelFormat gray_pix_fmts[] = {GRAY_PIXEL_FORMATS};
static const enum AVPixelFormat yuv_pix_fmts[]  = {YUV_PIXEL_FORMATS};
static const enum AVPixelFormat any_pix_fmts[]  = {RGB_PIXEL_FORMATS,
64 65 66
                                                 GRAY_PIXEL_FORMATS,
                                                 YUV_PIXEL_FORMATS};

67
typedef struct {
68
    AVClass *class;
69 70
    opj_dparameters_t dec_params;
    AVFrame image;
71 72
    int lowres;
    int lowqual;
73 74
} LibOpenJPEGContext;

75
static int libopenjpeg_matches_pix_fmt(const opj_image_t *img,
76
                                       enum AVPixelFormat pix_fmt)
77
{
78
    const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
79 80
    int match = 1;

81
    if (desc->nb_components != img->numcomps) {
82 83 84
        return 0;
    }

85
    switch (desc->nb_components) {
86 87
    case 4:
        match = match &&
88
            desc->comp[3].depth_minus1 + 1 >= img->comps[3].prec &&
89 90 91 92
            1 == img->comps[3].dx &&
            1 == img->comps[3].dy;
    case 3:
        match = match &&
93 94 95
            desc->comp[2].depth_minus1 + 1 >= img->comps[2].prec &&
            1 << desc->log2_chroma_w == img->comps[2].dx &&
            1 << desc->log2_chroma_h == img->comps[2].dy;
96 97
    case 2:
        match = match &&
98 99 100
            desc->comp[1].depth_minus1 + 1 >= img->comps[1].prec &&
            1 << desc->log2_chroma_w == img->comps[1].dx &&
            1 << desc->log2_chroma_h == img->comps[1].dy;
101 102
    case 1:
        match = match &&
103
            desc->comp[0].depth_minus1 + 1 >= img->comps[0].prec &&
104 105 106 107 108 109 110 111 112
            1 == img->comps[0].dx &&
            1 == img->comps[0].dy;
    default:
        break;
    }

    return match;
}

113
static enum AVPixelFormat libopenjpeg_guess_pix_fmt(const opj_image_t *image)
114 115
{
    int index;
116
    const enum AVPixelFormat *possible_fmts = NULL;
117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143
    int possible_fmts_nb = 0;

    switch (image->color_space) {
    case CLRSPC_SRGB:
        possible_fmts = rgb_pix_fmts;
        possible_fmts_nb = FF_ARRAY_ELEMS(rgb_pix_fmts);
        break;
    case CLRSPC_GRAY:
        possible_fmts = gray_pix_fmts;
        possible_fmts_nb = FF_ARRAY_ELEMS(gray_pix_fmts);
        break;
    case CLRSPC_SYCC:
        possible_fmts = yuv_pix_fmts;
        possible_fmts_nb = FF_ARRAY_ELEMS(yuv_pix_fmts);
        break;
    default:
        possible_fmts = any_pix_fmts;
        possible_fmts_nb = FF_ARRAY_ELEMS(any_pix_fmts);
        break;
    }

    for (index = 0; index < possible_fmts_nb; ++index) {
        if (libopenjpeg_matches_pix_fmt(image, possible_fmts[index])) {
            return possible_fmts[index];
        }
    }

144
    return AV_PIX_FMT_NONE;
145 146
}

147
static inline int libopenjpeg_ispacked(enum AVPixelFormat pix_fmt)
148
{
149
    const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
150 151
    int i, component_plane;

152
    if (pix_fmt == AV_PIX_FMT_GRAY16)
153 154
        return 0;

155 156 157
    component_plane = desc->comp[0].plane;
    for (i = 1; i < desc->nb_components; i++) {
        if (component_plane != desc->comp[i].plane)
158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218
            return 0;
    }
    return 1;
}

static void libopenjpeg_copy_to_packed8(AVFrame *picture, opj_image_t *image)
{
    uint8_t *img_ptr;
    int index, x, y, c;

    for (y = 0; y < picture->height; y++) {
        index = y*picture->width;
        img_ptr = picture->data[0] + y*picture->linesize[0];
        for (x = 0; x < picture->width; x++, index++) {
            for (c = 0; c < image->numcomps; c++) {
                *img_ptr++ = image->comps[c].data[index];
            }
        }
    }
}

static void libopenjpeg_copy_to_packed16(AVFrame *picture, opj_image_t *image)
{
    uint16_t *img_ptr;
    int index, x, y, c;
    int adjust[4];

    for (x = 0; x < image->numcomps; x++)
        adjust[x] = FFMAX(FFMIN(16 - image->comps[x].prec, 8), 0);

    for (y = 0; y < picture->height; y++) {
        index = y*picture->width;
        img_ptr = (uint16_t*) (picture->data[0] + y*picture->linesize[0]);
        for (x = 0; x < picture->width; x++, index++) {
            for (c = 0; c < image->numcomps; c++) {
                *img_ptr++ = image->comps[c].data[index] << adjust[c];
            }
        }
    }
}

static void libopenjpeg_copyto8(AVFrame *picture, opj_image_t *image)
{
    int *comp_data;
    uint8_t *img_ptr;
    int index, x, y;

    for (index = 0; index < image->numcomps; index++) {
        comp_data = image->comps[index].data;
        for (y = 0; y < image->comps[index].h; y++) {
            img_ptr = picture->data[index] + y * picture->linesize[index];
            for (x = 0; x < image->comps[index].w; x++) {
                *img_ptr = (uint8_t) *comp_data;
                img_ptr++;
                comp_data++;
            }
        }
    }
}

static void libopenjpeg_copyto16(AVFrame *p, opj_image_t *image)
219
{
220 221 222 223 224 225 226 227 228 229 230 231 232 233 234
    int *comp_data;
    uint16_t *img_ptr;
    int index, x, y;

    for (index = 0; index < image->numcomps; index++) {
        comp_data = image->comps[index].data;
        for (y = 0; y < image->comps[index].h; y++) {
            img_ptr = (uint16_t*) (p->data[index] + y * p->linesize[index]);
            for (x = 0; x < image->comps[index].w; x++) {
                *img_ptr = *comp_data;
                img_ptr++;
                comp_data++;
            }
        }
    }
235 236 237 238 239 240 241
}

static av_cold int libopenjpeg_decode_init(AVCodecContext *avctx)
{
    LibOpenJPEGContext *ctx = avctx->priv_data;

    opj_set_default_decoder_parameters(&ctx->dec_params);
242
    avcodec_get_frame_defaults(&ctx->image);
243 244 245 246
    avctx->coded_frame = &ctx->image;
    return 0;
}

247 248 249 250 251 252 253 254
static av_cold int libopenjpeg_decode_init_thread_copy(AVCodecContext *avctx)
{
    LibOpenJPEGContext *ctx = avctx->priv_data;

    avctx->coded_frame = &ctx->image;
    return 0;
}

255 256
static int libopenjpeg_decode_frame(AVCodecContext *avctx,
                                    void *data, int *data_size,
257
                                    AVPacket *avpkt)
258
{
259
    uint8_t *buf = avpkt->data;
260
    int buf_size = avpkt->size;
261 262
    LibOpenJPEGContext *ctx = avctx->priv_data;
    AVFrame *picture = &ctx->image, *output = data;
263
    const AVPixFmtDescriptor *desc;
264 265 266
    opj_dinfo_t *dec;
    opj_cio_t *stream;
    opj_image_t *image;
267 268 269 270
    int width, height, ret = -1;
    int pixel_size = 0;
    int ispacked = 0;
    int i;
271 272 273 274

    *data_size = 0;

    // Check if input is a raw jpeg2k codestream or in jp2 wrapping
275 276 277
    if ((AV_RB32(buf)     == 12)           &&
        (AV_RB32(buf + 4) == JP2_SIG_TYPE) &&
        (AV_RB32(buf + 8) == JP2_SIG_VALUE)) {
Jai Menon's avatar
Jai Menon committed
278
        dec = opj_create_decompress(CODEC_JP2);
279
    } else {
280 281
        /* If the AVPacket contains a jp2c box, then skip to
         * the starting byte of the codestream. */
282 283
        if (AV_RB32(buf + 4) == AV_RB32("jp2c"))
            buf += 8;
Jai Menon's avatar
Jai Menon committed
284
        dec = opj_create_decompress(CODEC_J2K);
285 286
    }

287
    if (!dec) {
288 289 290 291 292
        av_log(avctx, AV_LOG_ERROR, "Error initializing decoder.\n");
        return -1;
    }
    opj_set_event_mgr((opj_common_ptr)dec, NULL, NULL);

293
    ctx->dec_params.cp_limit_decoding = LIMIT_TO_MAIN_HEADER;
294 295
    ctx->dec_params.cp_reduce         = ctx->lowres;
    ctx->dec_params.cp_layer          = ctx->lowqual;
296 297 298
    // Tie decoder with decoding parameters
    opj_setup_decoder(dec, &ctx->dec_params);
    stream = opj_cio_open((opj_common_ptr)dec, buf, buf_size);
299 300 301 302

    if (!stream) {
        av_log(avctx, AV_LOG_ERROR,
               "Codestream could not be opened for reading.\n");
303 304 305 306
        opj_destroy_decompress(dec);
        return -1;
    }

307
    // Decode the header only.
308 309
    image = opj_decode_with_info(dec, stream, NULL);
    opj_cio_close(stream);
310 311

    if (!image) {
312 313 314 315
        av_log(avctx, AV_LOG_ERROR, "Error decoding codestream.\n");
        opj_destroy_decompress(dec);
        return -1;
    }
316

317 318
    width  = image->x1 - image->x0;
    height = image->y1 - image->y0;
319 320

    if (ctx->lowres) {
321
        width  = (width  + (1 << ctx->lowres) - 1) >> ctx->lowres;
322 323 324
        height = (height + (1 << ctx->lowres) - 1) >> ctx->lowres;
    }

325 326 327
    if (av_image_check_size(width, height, 0, avctx) < 0) {
        av_log(avctx, AV_LOG_ERROR,
               "%dx%d dimension invalid.\n", width, height);
328 329
        goto done;
    }
330

331 332
    avcodec_set_dimensions(avctx, width, height);

333
    if (avctx->pix_fmt != AV_PIX_FMT_NONE)
334
        if (!libopenjpeg_matches_pix_fmt(image, avctx->pix_fmt))
335
            avctx->pix_fmt = AV_PIX_FMT_NONE;
336

337
    if (avctx->pix_fmt == AV_PIX_FMT_NONE)
338 339
        avctx->pix_fmt = libopenjpeg_guess_pix_fmt(image);

340
    if (avctx->pix_fmt == AV_PIX_FMT_NONE) {
341 342
        av_log(avctx, AV_LOG_ERROR, "Unable to determine pixel format\n");
        ret = AVERROR_INVALIDDATA;
343
        goto done;
344 345
    }

346 347 348 349
    for (i = 0; i < image->numcomps; i++)
        if (image->comps[i].prec > avctx->bits_per_raw_sample)
            avctx->bits_per_raw_sample = image->comps[i].prec;

350
    if (picture->data[0])
351 352
        ff_thread_release_buffer(avctx, picture);

353
    if (ff_thread_get_buffer(avctx, picture) < 0) {
354
        av_log(avctx, AV_LOG_ERROR, "ff_thread_get_buffer() failed\n");
355
        goto done;
356
    }
357

358
    ctx->dec_params.cp_limit_decoding = NO_LIMITATION;
359
    // Tie decoder with decoding parameters.
360 361
    opj_setup_decoder(dec, &ctx->dec_params);
    stream = opj_cio_open((opj_common_ptr)dec, buf, buf_size);
362 363 364
    if (!stream) {
        av_log(avctx, AV_LOG_ERROR,
               "Codestream could not be opened for reading.\n");
365
        goto done;
366 367
    }

368 369
    opj_image_destroy(image);
    // Decode the codestream
370 371 372
    image = opj_decode_with_info(dec, stream, NULL);
    opj_cio_close(stream);

373 374 375 376 377
    if (!image) {
        av_log(avctx, AV_LOG_ERROR, "Error decoding codestream.\n");
        goto done;
    }

378 379
    desc = av_pix_fmt_desc_get(avctx->pix_fmt);
    pixel_size = desc->comp[0].step_minus1 + 1;
380 381 382 383 384 385 386 387 388 389 390 391 392 393 394
    ispacked = libopenjpeg_ispacked(avctx->pix_fmt);

    switch (pixel_size) {
    case 1:
        if (ispacked) {
            libopenjpeg_copy_to_packed8(picture, image);
        } else {
            libopenjpeg_copyto8(picture, image);
        }
        break;
    case 2:
        if (ispacked) {
            libopenjpeg_copy_to_packed8(picture, image);
        } else {
            libopenjpeg_copyto16(picture, image);
395
        }
396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411
        break;
    case 3:
    case 4:
        if (ispacked) {
            libopenjpeg_copy_to_packed8(picture, image);
        }
        break;
    case 6:
    case 8:
        if (ispacked) {
            libopenjpeg_copy_to_packed16(picture, image);
        }
        break;
    default:
        av_log(avctx, AV_LOG_ERROR, "unsupported pixel size %d\n", pixel_size);
        goto done;
412 413 414 415
    }

    *output    = ctx->image;
    *data_size = sizeof(AVPicture);
416
    ret        = buf_size;
417 418 419 420 421 422 423 424 425 426 427

done:
    opj_image_destroy(image);
    opj_destroy_decompress(dec);
    return ret;
}

static av_cold int libopenjpeg_decode_close(AVCodecContext *avctx)
{
    LibOpenJPEGContext *ctx = avctx->priv_data;

428
    if (ctx->image.data[0])
429
        ff_thread_release_buffer(avctx, &ctx->image);
430
    return 0;
431 432
}

433 434 435 436
#define OFFSET(x) offsetof(LibOpenJPEGContext, x)
#define VD AV_OPT_FLAG_VIDEO_PARAM | AV_OPT_FLAG_DECODING_PARAM

static const AVOption options[] = {
437 438
    { "lowqual", "Limit the number of layers used for decoding",    OFFSET(lowqual), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, INT_MAX, VD },
    { "lowres",  "Lower the decoding resolution by a power of two", OFFSET(lowres),  AV_OPT_TYPE_INT, { .i64 = 0 }, 0, INT_MAX, VD },
439 440 441 442 443 444 445 446 447
    { NULL },
};

static const AVClass class = {
    .class_name = "libopenjpeg",
    .item_name  = av_default_item_name,
    .option     = options,
    .version    = LIBAVUTIL_VERSION_INT,
};
448

449
AVCodec ff_libopenjpeg_decoder = {
450 451
    .name             = "libopenjpeg",
    .type             = AVMEDIA_TYPE_VIDEO,
452
    .id               = AV_CODEC_ID_JPEG2000,
453 454 455 456 457
    .priv_data_size   = sizeof(LibOpenJPEGContext),
    .init             = libopenjpeg_decode_init,
    .close            = libopenjpeg_decode_close,
    .decode           = libopenjpeg_decode_frame,
    .capabilities     = CODEC_CAP_DR1 | CODEC_CAP_FRAME_THREADS,
458
    .long_name        = NULL_IF_CONFIG_SMALL("OpenJPEG JPEG 2000"),
459
    .priv_class       = &class,
460
    .init_thread_copy = ONLY_IF_THREADS_ENABLED(libopenjpeg_decode_init_thread_copy),
461
};