Kakadu Software is a comprehensive, heavily optimized, fully compliant software toolkit for JPEG2000 developers. It is a complete implementation of the JPEG2000 standard, Part 1 (i.e., ISO/IEC 15444-1) and Part 15 (ISO/IEC 15444-15, a.k.a. High Throughput JPEG 2000). The Software Development Kit (SDK) also provides a comprehensive implementation of most features from Part 2 of the JPEG2000 standard, including general multi-component transforms, arbitrary wavelet transform kernels, arbitrary decomposition structures and non-linear point transforms (especially useful for HDR and float/half-float compression).
The JPEG 2000 compression standard is much more sophisticated and versatile than the JPEG standard, both from a computational and a conceptual perspective.
Kakadu Software features:
- Fully supports region-of-interest, resolution-of-interest and component-of-interest access to all JPEG 2000 content, including extremely efficient (and low memory) rendering from vast image surfaces where random access pointers are available in the code-stream.
- Highly functional and efficient client and server components for the JPIP interactive image/video browsing standard, allowing responsive navigation through huge media, even over very slow or error prone networks.
- Extensive and convenient support for Java native interfaces.
- Automatically builds bindings for C# and Visual Basic programmers
- Multi-threaded processing to fully utilize parallel processing resources (multiple CPUs, multi-core CPUs or hyper-threading).
- A carefully engineered thread scheduler so once you have created a multi-threaded environment and populated it with one thread for each physical/virtual processor on your system, close to 100% utilization is achieved by using all computational resources.
- Highly optimized to efficiently utilize memory and CPU resources throughout the entire compression/decompression pipeline. In particular, when using the new High Throughput block coding algorithm from JPEG 2000 Part-15, Kakadu is almost certainly faster than any implementation of the original JPEG algorithm, even without taking into account the efficient region/resolution-of-interest access features of JPEG 2000 and Kakadu.