[FFmpeg-devel] [PATCH v4 1/3] avcodec/jpeg2000dec: Add support for CAP and CPF markers
WATANABE Osamu
owatanab at es.takushoku-u.ac.jp
Sat Jul 13 13:31:50 EEST 2024
> @Osamu, did you check if the decoded images fall within the tolerances
> specified in 15444-4?
Yes, I did. It has been confirmed that the decoder passes all the test cases defined in 15444-4.
The errors (PAE and MSE) in the reconstructed images are within the allowable tolerances.
> On Jul 12, 2024, at 13:57, Pierre-Anthony Lemieux <pal at sandflow.com> wrote:
>
> I checked that the patch results in FFMPEG decoding without error the
> additional conformance codestreams at [1].
>
> [1] https://ffmpeg.org/pipermail/ffmpeg-devel/2024-June/329727.html
>
> @Osamu, did you check if the decoded images fall within the tolerances
> specified in 15444-4?
>
> There is an errant space character at the end of four lines for some
> reason. I can fix that before merging.
>
> Any other comments? If not, I plan to merge sometime this week, and
> plan to then submit a patch to add the conformance codestream to FATE.
>
> On Mon, Jun 24, 2024 at 6:37?AM Osamu Watanabe
> <owatanab at es.takushoku-u.ac.jp> wrote:
>>
>> This commit adds support for CAP and CPF markers.
>>
>> The previous version
>> (v3, https://patchwork.ffmpeg.org/project/ffmpeg/list/?series=12246)
>> was wrongly separated. I have fixed the way to separation.
>>
>> The changes are essentially the same as v2
>> (https://patchwork.ffmpeg.org/project/ffmpeg/list/?series=12199).
>> The suggested modifications have been made according to the discussion
>> on this mailing list.
>>
>> Signed-off-by: Osamu Watanabe <owatanab at es.takushoku-u.ac.jp>
>> ---
>> libavcodec/jpeg2000.h | 8 +++
>> libavcodec/jpeg2000dec.c | 112 ++++++++++++++++++++++++++++++++++++++-
>> libavcodec/jpeg2000dec.h | 7 +++
>> 3 files changed, 126 insertions(+), 1 deletion(-)
>>
>> diff --git a/libavcodec/jpeg2000.h b/libavcodec/jpeg2000.h
>> index d004c08f10..4bdc38df7c 100644
>> --- a/libavcodec/jpeg2000.h
>> +++ b/libavcodec/jpeg2000.h
>> @@ -37,12 +37,14 @@
>>
>> enum Jpeg2000Markers {
>> JPEG2000_SOC = 0xff4f, // start of codestream
>> + JPEG2000_CAP = 0xff50, // extended capabilities
>> JPEG2000_SIZ = 0xff51, // image and tile size
>> JPEG2000_COD, // coding style default
>> JPEG2000_COC, // coding style component
>> JPEG2000_TLM = 0xff55, // tile-part length, main header
>> JPEG2000_PLM = 0xff57, // packet length, main header
>> JPEG2000_PLT, // packet length, tile-part header
>> + JPEG2000_CPF, // corresponding profile
>> JPEG2000_QCD = 0xff5c, // quantization default
>> JPEG2000_QCC, // quantization component
>> JPEG2000_RGN, // region of interest
>> @@ -58,6 +60,12 @@ enum Jpeg2000Markers {
>> JPEG2000_EOC = 0xffd9, // end of codestream
>> };
>>
>> +enum JPEG2000_Ccap15_b14_15_params {
>> + HTJ2K_HTONLY = 0, // HTONLY, bit 14 and 15 are 0
>> + HTJ2K_HTDECLARED, // HTDECLARED, bit 14 = 1 and bit 15 = 0
>> + HTJ2K_MIXED = 3, // MIXED, bit 14 and 15 are 1
>> +};
>> +
>> #define JPEG2000_SOP_FIXED_BYTES 0xFF910004
>> #define JPEG2000_SOP_BYTE_LENGTH 6
>>
>> diff --git a/libavcodec/jpeg2000dec.c b/libavcodec/jpeg2000dec.c
>> index 091931b1ff..d1046661c4 100644
>> --- a/libavcodec/jpeg2000dec.c
>> +++ b/libavcodec/jpeg2000dec.c
>> @@ -408,6 +408,73 @@ static int get_siz(Jpeg2000DecoderContext *s)
>> s->avctx->bits_per_raw_sample = s->precision;
>> return 0;
>> }
>> +/* get extended capabilities (CAP) marker segment */
>> +static int get_cap(Jpeg2000DecoderContext *s, Jpeg2000CodingStyle *c)
>> +{
>> + uint32_t Pcap;
>> + uint16_t Ccap_i[32] = { 0 };
>> + uint16_t Ccap_15;
>> + uint8_t P;
>> +
>> + if (bytestream2_get_bytes_left(&s->g) < 6) {
>> + av_log(s->avctx, AV_LOG_ERROR, "Insufficient space for CAP\n");
>> + return AVERROR_INVALIDDATA;
>> + }
>> +
>> + Pcap = bytestream2_get_be32u(&s->g);
>> + s->isHT = (Pcap >> (31 - (15 - 1))) & 1;
>> + for (int i = 0; i < 32; i++) {
>> + if ((Pcap >> (31 - i)) & 1)
>> + Ccap_i[i] = bytestream2_get_be16u(&s->g);
>> + }
>> + Ccap_15 = Ccap_i[14];
>> + if (s->isHT == 1) {
>> + av_log(s->avctx, AV_LOG_INFO, "This is an HTJ2K codestream.\n");
>> + // Bits 14-15
>> + switch ((Ccap_15 >> 14) & 0x3) {
>> + case 0x3:
>> + s->Ccap15_b14_15 = HTJ2K_MIXED;
>> + break;
>> + case 0x1:
>> + s->Ccap15_b14_15 = HTJ2K_HTDECLARED;
>> + break;
>> + case 0x0:
>> + s->Ccap15_b14_15 = HTJ2K_HTONLY;
>> + break;
>> + default:
>> + av_log(s->avctx, AV_LOG_ERROR, "Unknown CCap value.\n");
>> + return AVERROR(EINVAL);
>> + break;
>> + }
>> + // Bit 13
>> + if ((Ccap_15 >> 13) & 1) {
>> + av_log(s->avctx, AV_LOG_ERROR, "MULTIHT set is not supported.\n");
>> + return AVERROR_PATCHWELCOME;
>> + }
>> + // Bit 12
>> + s->Ccap15_b12 = (Ccap_15 >> 12) & 1;
>> + // Bit 11
>> + s->Ccap15_b11 = (Ccap_15 >> 11) & 1;
>> + // Bit 5
>> + s->Ccap15_b05 = (Ccap_15 >> 5) & 1;
>> + // Bit 0-4
>> + P = Ccap_15 & 0x1F;
>> + if (!P)
>> + s->HT_MAGB = 8;
>> + else if (P < 20)
>> + s->HT_MAGB = P + 8;
>> + else if (P < 31)
>> + s->HT_MAGB = 4 * (P - 19) + 27;
>> + else
>> + s->HT_MAGB = 74;
>> +
>> + if (s->HT_MAGB > 31) {
>> + av_log(s->avctx, AV_LOG_ERROR, "Available internal precision is exceeded (MAGB> 31).\n");
>> + return AVERROR_PATCHWELCOME;
>> + }
>> + }
>> + return 0;
>> +}
>>
>> /* get common part for COD and COC segments */
>> static int get_cox(Jpeg2000DecoderContext *s, Jpeg2000CodingStyle *c)
>> @@ -802,6 +869,15 @@ static int read_crg(Jpeg2000DecoderContext *s, int n)
>> bytestream2_skip(&s->g, n - 2);
>> return 0;
>> }
>> +
>> +static int read_cpf(Jpeg2000DecoderContext *s, int n)
>> +{
>> + if (bytestream2_get_bytes_left(&s->g) < (n - 2))
>> + return AVERROR_INVALIDDATA;
>> + bytestream2_skip(&s->g, n - 2);
>> + return 0;
>> +}
>> +
>> /* Tile-part lengths: see ISO 15444-1:2002, section A.7.1
>> * Used to know the number of tile parts and lengths.
>> * There may be multiple TLMs in the header.
>> @@ -965,6 +1041,14 @@ static int init_tile(Jpeg2000DecoderContext *s, int tileno)
>> comp->roi_shift = s->roi_shift[compno];
>> if (!codsty->init)
>> return AVERROR_INVALIDDATA;
>> + if (s->isHT && (!s->Ccap15_b05) && (!codsty->transform)) {
>> + av_log(s->avctx, AV_LOG_ERROR, "Transformation = 0 (lossy DWT) is found in HTREV HT set\n");
>> + return AVERROR_INVALIDDATA;
>> + }
>> + if (s->isHT && s->Ccap15_b14_15 != (codsty->cblk_style >> 6) && s->Ccap15_b14_15 != HTJ2K_HTONLY) {
>> + av_log(s->avctx, AV_LOG_ERROR, "SPcod/SPcoc value does not match bit 14-15 values of Ccap15\n");
>> + return AVERROR_INVALIDDATA;
>> + }
>> if (ret = ff_jpeg2000_init_component(comp, codsty, qntsty,
>> s->cbps[compno], s->cdx[compno],
>> s->cdy[compno], s->avctx))
>> @@ -2187,22 +2271,43 @@ static int jpeg2000_read_main_headers(Jpeg2000DecoderContext *s)
>> if (!s->tile)
>> s->numXtiles = s->numYtiles = 0;
>> break;
>> + case JPEG2000_CAP:
>> + if (!s->ncomponents) {
>> + av_log(s->avctx, AV_LOG_ERROR, "CAP marker segment shall come after SIZ\n");
>> + return AVERROR_INVALIDDATA;
>> + }
>> + ret = get_cap(s, codsty);
>> + break;
>> case JPEG2000_COC:
>> + if (s->in_tile_headers == 1 && s->isHT && (!s->Ccap15_b11))
>> + av_log(s->avctx, AV_LOG_WARNING, "COC marker is found in HOMOGENEOUS HT set\n");
>> ret = get_coc(s, codsty, properties);
>> break;
>> case JPEG2000_COD:
>> + if (s->in_tile_headers == 1 && s->isHT && (!s->Ccap15_b11))
>> + av_log(s->avctx, AV_LOG_WARNING, "COD marker is found in HOMOGENEOUS HT set\n");
>> ret = get_cod(s, codsty, properties);
>> break;
>> case JPEG2000_RGN:
>> + if (s->in_tile_headers == 1 && s->isHT && (!s->Ccap15_b11))
>> + av_log(s->avctx, AV_LOG_WARNING, "RGB marker is found in HOMOGENEOUS HT set\n");
>> ret = get_rgn(s, len);
>> + if ((!s->Ccap15_b12) && s->isHT)
>> + av_log(s->avctx, AV_LOG_WARNING, "RGN marker is found in RGNFREE HT set\n");
>> break;
>> case JPEG2000_QCC:
>> + if (s->in_tile_headers == 1 && s->isHT && (!s->Ccap15_b11))
>> + av_log(s->avctx, AV_LOG_WARNING, "QCC marker is found in HOMOGENEOUS HT set\n");
>> ret = get_qcc(s, len, qntsty, properties);
>> break;
>> case JPEG2000_QCD:
>> + if (s->in_tile_headers == 1 && s->isHT && (!s->Ccap15_b11))
>> + av_log(s->avctx, AV_LOG_WARNING, "QCD marker is found in HOMOGENEOUS HT set\n");
>> ret = get_qcd(s, len, qntsty, properties);
>> break;
>> case JPEG2000_POC:
>> + if (s->in_tile_headers == 1 && s->isHT && (!s->Ccap15_b11))
>> + av_log(s->avctx, AV_LOG_WARNING, "POC marker is found in HOMOGENEOUS HT set\n");
>> ret = get_poc(s, len, poc);
>> break;
>> case JPEG2000_SOT:
>> @@ -2252,9 +2357,14 @@ static int jpeg2000_read_main_headers(Jpeg2000DecoderContext *s)
>> "Cannot have both PPT and PPM marker.\n");
>> return AVERROR_INVALIDDATA;
>> }
>> -
>> + if ((!s->Ccap15_b11) && s->isHT)
>> + av_log(s->avctx, AV_LOG_WARNING, "PPT marker is found in HOMOGENEOUS HT set\n");
>> ret = get_ppt(s, len);
>> break;
>> + case JPEG2000_CPF:
>> + // Corresponding profile marker
>> + ret = read_cpf(s, len);
>> + break;
>> default:
>> av_log(s->avctx, AV_LOG_ERROR,
>> "unsupported marker 0x%.4"PRIX16" at pos 0x%X\n",
>> diff --git a/libavcodec/jpeg2000dec.h b/libavcodec/jpeg2000dec.h
>> index d0ca6e7a79..326a572722 100644
>> --- a/libavcodec/jpeg2000dec.h
>> +++ b/libavcodec/jpeg2000dec.h
>> @@ -112,6 +112,13 @@ typedef struct Jpeg2000DecoderContext {
>> Jpeg2000Tile *tile;
>> Jpeg2000DSPContext dsp;
>>
>> + uint8_t isHT; // HTJ2K?
>> + uint8_t Ccap15_b14_15; // HTONLY(= 0) or HTDECLARED(= 1) or MIXED(= 3) ?
>> + uint8_t Ccap15_b12; // RGNFREE(= 0) or RGN(= 1)?
>> + uint8_t Ccap15_b11; // HOMOGENEOUS(= 0) or HETEROGENEOUS(= 1) ?
>> + uint8_t Ccap15_b05; // HTREV(= 0) or HTIRV(= 1) ?
>> + uint8_t HT_MAGB; // MAGB value
>> +
>> /*options parameters*/
>> int reduction_factor;
>> } Jpeg2000DecoderContext;
>> --
>> 2.34.1
>>
> _______________________________________________
> ffmpeg-devel mailing list
> ffmpeg-devel at ffmpeg.org
> https://ffmpeg.org/mailman/listinfo/ffmpeg-devel
>
> To unsubscribe, visit link above, or email
> ffmpeg-devel-request at ffmpeg.org with subject "unsubscribe".
More information about the ffmpeg-devel
mailing list