Skip to content

cgohlke/imagecodecs

BranchesTags

Repository files navigation

Image transformation, compression, and decompression codecs

Imagecodecs is a Python library that provides block-oriented, in-memory buffer transformation, compression, and decompression functions for use in Tifffile, Czifile, Zarr, kerchunk, and other scientific image input/output packages.

Decode and/or encode functions are implemented for Zlib (DEFLATE), GZIP, LZMA, ZStandard (ZSTD), Blosc, Brotli, Snappy, BZ2, LZ4, LZ4F, LZ4HC, LZ4H5, LZW, LZO, LZF, LZFSE, LZHAM, PGLZ (PostgreSQL LZ), RCOMP (Rice), ZFP, SZ3, Pcodec, SPERR, AEC, SZIP, LERC, EER, NPY, BCn, DDS, BMP, PNG, APNG, GIF, TIFF, WebP, JPEG 8 and 12-bit, Lossless JPEG (LJPEG, LJ92, JPEGLL), JPEG 2000 (JP2, J2K), JPEG LS, JPEG XL, JPEG XS, JPEG XR (WDP, HD Photo), Ultra HDR (JPEG_R), MOZJPEG, AVIF, HEIF, QOI, RGBE (HDR), Jetraw, DICOMRLE, PackBits, Packed Integers, Delta, XOR Delta, Floating Point Predictor, Bitorder reversal, Byteshuffle, Bitshuffle, Float24 (24-bit floating point), Quantize (Scale, BitGroom, BitRound, GranularBR), and CMS (color space transformations). Checksum functions are implemented for crc32, adler32, fletcher32, and Jenkins lookup3.

Author:Christoph Gohlke
License:BSD 3-Clause
Version:2024.9.22
DOI:10.5281/zenodo.6915978

Quickstart

Install the imagecodecs package and all dependencies from the Python Package Index:

python -m pip install -U "imagecodecs[all]"

Imagecodecs is also available in other package repositories such as Anaconda, MSYS2, and MacPorts.

See Requirements and Notes for building from source.

See Examples for using the programming interface.

Source code and support are available on GitHub.

Requirements

This revision was tested with the following requirements and dependencies (other versions may work):

  • CPython 3.10.11, 3.11.9, 3.12.6, 3.13.0rc2 64-bit
  • Numpy 2.1.1
  • numcodecs 0.13.0 (optional, for Zarr compatible codecs)

Build requirements:

Vendored requirements:

Test requirements:

Revisions

2024.9.22

  • Pass 7644 tests.
  • Use libjpeg-turbo for all Lossless JPEG bit-depths if possible (#105).
  • Fix PackBits encoder fails to skip short replication blocks (#107).
  • Fix JPEG2K encoder leaving trailing random bytes (#104).
  • Fix encoding and decoding JPEG XL with custom bitspersample (#102).
  • Improve error handling in lzf_decode (#103).
  • Add Ultra HDR (JPEG_R) codec based on libultrahdr library (#108).
  • Add JPEGXS codec based on libjxs library (source only).
  • Add SZ3 codec based on SZ3 library.
  • Deprecate Python 3.9, support Python 3.13.

2024.6.1

  • Fix segfault in sperr_decode.
  • Fix segfault when strided-decoding into buffers with unexpected shapes (#98).
  • Fix jpeg2k_encoder output buffer too small (#101).
  • Add PCODEC codec based on pcodec library.
  • Support NumPy 2.

2024.1.1

  • Add 8/24-bit BMP codec.
  • Add SPERR codec based on SPERR library.
  • Add LZO decoder based on lzokay library.
  • Add DICOMRLE decoder.
  • Enable float16 in CMS codec.
  • Enable MCT for lossless JPEG2K encoder (#88).
  • Ignore pad-byte in PackBits decoder (#86).
  • Fix heif_write_callback error message not set.
  • Require lcms2 2.16 with issue-420 fixes.
  • Require libjxl 0.9, libaec 1.1, Cython 3.

2023.9.18

  • Rebuild with updated dependencies fixes CVE-2024-4863.

2023.9.4

  • Map avif_encode level parameter to quality (breaking).
  • Support monochrome images in avif_encode.
  • Add numthreads parameter to avif_decode (fix imread of AVIF).
  • Add quantize filter (BitGroom, BitRound, GBR) via nc4var.c.
  • Add LZ4H5 codec.
  • Support more BCn compressed DDS fourcc types.
  • Require libavif 1.0.

2023.8.12

  • Add EER (Electron Event Representation) decoder.
  • Add option to pass initial value to crc32 and adler32 checksum functions.
  • Add fletcher32 and lookup3 checksum functions via HDF5's h5checksum.c.
  • Add Checksum codec for numcodecs.

2023.7.10

  • Rebuild with optimized compile flags.

2023.7.4

  • Add BCn and DDS decoder via bcdec library.
  • Add functions to transcode JPEG XL to/from JPEG (#78).
  • Add option to decode select frames from animated WebP.
  • Use legacy JPEG8 codec when building without libjpeg-turbo 3 (#65).
  • Change blosc2_encode defaults to match blosc2-python (breaking).
  • Fix segfault writing JPEG2K with more than 4 samples.
  • Fix some codecs returning bytearray by default.
  • Fully vendor cfitsio's ricecomp.c.
  • Drop support for Python 3.8 and numpy < 1.21 (NEP29).
  • …

Refer to the CHANGES file for older revisions.

Objectives

Many scientific image storage formats like TIFF, CZI, DICOM, HDF, and Zarr are containers that hold large numbers of small data segments (chunks, tiles, stripes), which are encoded using a variety of compression and pre-filtering methods. Metadata common to all data segments are typically stored separate from the segments.

The purpose of the Imagecodecs library is to support Python modules in encoding and decoding such data segments. The specific aims are:

  • Provide functions for encoding and decoding small image data segments in-memory (not in-file) from and to bytes or numpy arrays for many compression and filtering methods.
  • Support image formats and compression methods not available elsewhere in the Python ecosystem.
  • Reduce the runtime dependency on numerous, large, inapt, or unmaintained Python packages. The imagecodecs package only depends on numpy.
  • Implement codecs as Cython wrappers of 3rd party libraries with a C API and permissive license if exists, else use own C library. Provide Cython definition files for the wrapped C libraries.
  • Release the Python global interpreter lock (GIL) during extended native/C function calls for multi-threaded use.

Accessing parts of large data segments and reading metadata from segments are out of the scope of this library.

Notes

This library is largely a work in progress.

The API is not stable yet and might change between revisions.

Python <= 3.8 is no longer supported. 32-bit versions are deprecated.

Works on little-endian platforms only.

Supported platforms are win_amd64, win_arm64, win32, macosx_x86_64, macosx_arm64, manylinux_x86_64, and manylinux_aarch64.

Wheels may not be available for all platforms and all releases.

Only the win_amd64 wheels include all features.

The tiff, bcn, dds, dicomrle, eer, lzo, packints, and jpegsof3 codecs are currently decode-only.

The heif, jetraw, and jpegxs codecs are distributed as source code only due to license and possible patent usage issues.

The latest Microsoft Visual C++ Redistributable for Visual Studio 2015-2022 is required on Windows.

Refer to the imagecodecs/licenses folder for 3rd-party library licenses.

This software is based in part on the work of the Independent JPEG Group.

Update pip and setuptools to the latest version before installing imagecodecs:

python -m pip install -U pip setuptools wheel Cython

Before building imagecodecs from source code, install required tools and libraries. For example, on latest Ubuntu Linux distributions:

sudo apt-get install build-essential python3-dev cython3 python3-pip python3-setuptools python3-wheel python3-numpy libdeflate-dev libjpeg-dev libjxr-dev liblcms2-dev liblz4-dev liblerc-dev liblzma-dev libopenjp2-7-dev libpng-dev libtiff-dev libwebp-dev libz-dev libzstd-dev

To build and install imagecodecs from source code, run:

python -m pip install .

Many extensions are disabled by default when building from source.

To define which extensions are built, or to modify build settings such as library names and compiler arguments, provide a imagecodecs_distributor_setup.customize_build function, which is imported and executed during setup. See setup.py for pre-defined customize_build functions.

Other projects providing imaging or compression codecs: Python zlib, Python bz2, Python lzma, backports.lzma, python-lzo, python-lzw, python-lerc, wavpack-numcodecs, packbits, isa-l.igzip, fpzip, libmng, OpenEXR (EXR, PIZ, PXR24, B44, DWA), pyJetraw, tinyexr, pytinyexr, pyroexr, JasPer, libjpeg (GPL), pylibjpeg, pylibjpeg-libjpeg (GPL), pylibjpeg-openjpeg, pylibjpeg-rle, glymur, pyheif, pyrus-cramjam, PyLZHAM, BriefLZ, QuickLZ (GPL), LZO (GPL), nvJPEG, nvJPEG2K, PyTurboJPEG, CCSDS123, LPC-Rice, CompressionAlgorithms, Compressonator, Wuffs, TinyDNG, OpenJPH, Grok (AGPL), MAFISC, B3D, fo-dicom.Codecs, jpegli.

Examples

Import the JPEG2K codec:

>>> from imagecodecs import (
...     jpeg2k_encode,
...     jpeg2k_decode,
...     jpeg2k_check,
...     jpeg2k_version,
...     JPEG2K,
... )

Check that the JPEG2K codec is available in the imagecodecs build:

>>> JPEG2K.available
True

Print the version of the JPEG2K codec's underlying OpenJPEG library:

>>> jpeg2k_version()
'openjpeg 2.5.2'

Encode a numpy array in lossless JP2 format:

>>> array = numpy.random.randint(100, 200, (256, 256, 3), numpy.uint8)
>>> encoded = jpeg2k_encode(array, level=0)
>>> bytes(encoded[:12])
b'\x00\x00\x00\x0cjP  \r\n\x87\n'

Check that the encoded bytes likely contain a JPEG 2000 stream:

>>> jpeg2k_check(encoded)
True

Decode the JP2 encoded bytes to a numpy array:

>>> decoded = jpeg2k_decode(encoded)
>>> numpy.array_equal(decoded, array)
True

Decode the JP2 encoded bytes to an existing numpy array:

>>> out = numpy.empty_like(array)
>>> _ = jpeg2k_decode(encoded, out=out)
>>> numpy.array_equal(out, array)
True

Not all codecs are fully implemented, raising exceptions at runtime:

>>> from imagecodecs import tiff_encode
>>> tiff_encode(array)
Traceback (most recent call last):
 ...
NotImplementedError: tiff_encode

Write the numpy array to a JP2 file:

>>> from imagecodecs import imwrite, imread
>>> imwrite('_test.jp2', array)

Read the image from the JP2 file as numpy array:

>>> image = imread('_test.jp2')
>>> numpy.array_equal(image, array)
True

Create a JPEG 2000 compressed Zarr array:

>>> import zarr
>>> import numcodecs
>>> from imagecodecs.numcodecs import Jpeg2k
>>> numcodecs.register_codec(Jpeg2k)
>>> zarr.zeros(
...     (4, 5, 512, 512, 3),
...     chunks=(1, 1, 256, 256, 3),
...     dtype='u1',
...     compressor=Jpeg2k(),
... )
<zarr.core.Array (4, 5, 512, 512, 3) uint8>

Access image data in a sequence of JP2 files via tifffile.FileSequence and dask.array:

>>> import tifffile
>>> import dask.array
>>> def jp2_read(filename):
...     with open(filename, 'rb') as fh:
...         data = fh.read()
...     return jpeg2k_decode(data)
...
>>> with tifffile.FileSequence(jp2_read, '*.jp2') as ims:
...     with ims.aszarr() as store:
...         dask.array.from_zarr(store)
...
dask.array<from-zarr, shape=(1, 256, 256, 3)...chunksize=(1, 256, 256, 3)...

Write the Zarr store to a fsspec ReferenceFileSystem in JSON format and open it as a Zarr array:

>>> store.write_fsspec(
...     'temp.json', url='file://', codec_id='imagecodecs_jpeg2k'
... )
>>> import fsspec
>>> mapper = fsspec.get_mapper(
...     'reference://', fo='temp.json', target_protocol='file'
... )
>>> zarr.open(mapper, mode='r')
<zarr.core.Array (1, 256, 256, 3) uint8 read-only>

View the image in the JP2 file from the command line:

python -m imagecodecs _test.jp2

About

Image transformation, compression, and decompression codecs

pypi.org/project/imagecodecs

Topics

python image png jpeg tiff webp lz4 lzma jpeg2000 lzw lzf packbits lerc jpegxr jpeg-ls zarr avif zfp jpegxl ultrahdr

Resources

Readme

License

BSD-3-Clause license
Activity

Stars

122 stars

Watchers

4 watching

Forks

24 forks
Report repository

Releases 16

v2024.9.22 Latest
Sep 21, 2024
+ 15 releases

Languages