ImageZero

Fast Lossless Color Image Compression

Introduction

ImageZero ("IZ") is a low-level, high-performance, in-memory, lossless color image compression C++ template library, available under BSD 2-clause license. It is intended to be used by image database applications, photo manipulation software, or other applications where simple RLE or no compression had been used because of performance or complexity considerations.

IZ is designed with photographic images in mind, where other fast compression algorithms (e.g. QuickLZ) fail to obtain any significant compression rate. With IZ, the compressed size for natural 24-bit photographic images is usually around 30% ... 50% of uncompressed size. Depending on the image, IZ often compresses better than PNG or JPEG-LS. For very compressible images, however, the compression rate can be much worse than the rate obtained with any other compression algorithm.

IZ is a symmetric coder, meaning both compression as well as decompression need about the same amount of time. On modern processors, the single-core performance is up to 180 megabytes per second of 24-bit RGB pixel data.

IZ is optimized for modern processors, featuring small cache-friendly Huffman coding tables and branchless implementations of nearly all operations. IZ takes advantage of 64-bit architecture improvements (e.g. MMX registers), but scales well to smaller or larger architectures.

Not yet: IZ has a flexible API, allowing to customize pixel accessors for any desired pixel color component layout (including alpha components), pixel iterators for different image storage layouts, and pixel predictors for different image types. The template library is self-contained, and does not dependent on any other library.

Note: IZ is early work in progress. The stream data format and the library API are not yet specified, and are subject to change without further notice. IZ does not specify or propose a container file format.


Downloads

PPM ⇔ IZ Test Program

Linux Binaries: http://download.opensuse.org/repositories/home:/cfeck/

Windows Binaries: iz32.exe iz64.exe

Useful Links

Forum: http://encode.ru/threads/1471-iz-New-fast-lossless-RGB-photo-compression

Blog: http://kdepepo.wordpress.com/

Source Repository: https://gitorious.org/imagezero

Today's git master: https://gitorious.org/imagezero/imagezero/archive-tarball/master

Independent Benchmarks:

Modified: 2013-09-06 by Christoph Feck

Valid XHTML 1.0 Strict

STudyTraY.de Banner Webspace for ImageZero kindly
sponsored by BMS-Shop