OpenHTJ2K

OpenHTJ2K is an open source implementation of ITU-T Rec.814 | ISO/IEC 15444-15 (a.k.a. JPEG 2000 Part 15, High-Throughput JPEG 2000; HTJ2K)

What OpenHTJ2K provides

OpenHTJ2K provides a shared liberary and sample applications having the following functionalities:

• Decoding of ITU-T Rec.800 | ISO/IEC 15444-1 (JPEG 2000 Part 1) or ITU-T Rec.814 | ISO/IEC 15444-15 (JPEG 2000 Part 15.) compliant codestreams
  • fully compliant with conformance testing defined in ITU-T Rec.803 | ISO 15444-4.
• Encoding an image into a codestream/JPH file which is compliant with HTJ2K
  • currently supports only HTJ2K. The optional markers like COC, POC, etc. are not implemented.
  • encoding of HT SigProp and MagRef passes are not implemented.
  • up to 12 bit per component sample is currently supported.
  • Quality control for lossy compression with Qfactor feature

Requirements

cmake (version 3.14 or later) and C++11 compliant compiler.

Building

Type the following command. ./ is a root of cloned repository and ${BUILD_DIR} is a build directory (for example, ../build or ./build and so on)

• You can also specify -DCMAKE_BUILD_TYPE=Debug or -DCMAKE_BUILD_TYPE=RelWithDebInfo to build with debug information.
• You can also specify -G "Xcode" to create a project for Xcode.
• You can also specify -G "Visual Studio 17 2022" to create a project for Visual Studio 2022. For the older versions, see https://cmake.org/cmake/help/latest/manual/cmake-generators.7.html#visual-studio-generators

cd ./
cmake -G "Unix Makefiles" -B ${BUILD_DIR} -DCMAKE_BUILD_TYPE=Release
cd  ${BUILD_DIR}
make

Then the executables should be found in ${BUILD_DIR}/bin directory.

Usage

Encoder

Only Part 15 compliant encoding is supported. Both .j2c (codestream) and .jph (file format) are available.

./open_htj2k_enc -i inputimage(in PNM/PGX/(TIFF, libtiff required) format) -o output [options...]

The encoder can take comma-separated multiple files. For example, components in YCbCr color space can be encoded by

./open_htj2k_enc -i inputY.pgx,inputCb.pgx,inputCr.pgx -o output 

options

• Stiles=Size
  • Size of tile "{height, width}", Default is equal to the image size
• Sorigin=Size
  • origin of the input image in the reference grid, Default is {0,0}
• Stile_origin=Size
  • origin of tiles in the reference grid, Default is {0,0}
• Clevels=Int
  • Valid range for number of DWT levels is from 0 to 32 (Default is 5)
• Creversible=yes or no
  • yes for lossless mode, no for lossy mode, Default is no
• Cblk=Size
  • Code-block size, Default is 64x64
• Cprecincts=Size
  • Precinct size
• Cycc=yes or no
  • yes to use RGB->YCbCr, Default is yes
• Corder
  • Progression order: LRCP, RLCP, RPCL, PCRL, CPRL, Default is LRCP
• Cuse_sop=yes or no
  • Default is no
• Cuse_eph=yes or no
  • Default is no
• Qstep=Float
  • 0.0 < base step size <= 2.0
• Qguard=Int
  • 0 to 7 for the number of guard bits, Default is 1
• Qderived=yes or no
  • yes switches the quantyzation style to derived (Default is no)
• Qfactor=Int
  • 0 to 100 for the quality of the lossy compressed image
  • for YCbCr inputs, valid chroma subsampling formats are 4:4:4, 4:2:0, and 4:2:2
• -jph_color_space
  • Color space of input components: RGB, YCC
  • if inputs are represented in YCbCr, use YCC
• -num_threads Int
  • number of threads to use in encode or decode
  • 0, which is the default, indicates usage of all threads

Decoder

The both Part 1 and Part 15 compliant decoding are supported.

./open_htj2k_dec -i codestream -o outputimage [-reduce n: number of DWT level reduction]

To see a help, use -h option.

Supported file types

Encoder

• input image formats: .pgm, .ppm, .pgx, .tif (libtiff required)
• output codestreams: .jhc (Part 15 codestream, .j2k and .j2c can be used as aliases), .jph (Part 15 file format)
  • Note: Specifying .jph as the output triggers a JPH file creation, otherwise just a codestream will be generated.

Decoder

• input codestreams : .j2k, .j2c, .jhc
• output image formats: .raw, .ppm, .pgm, .pgx