Jetraw Core

Jetraw is a raw image compressor designed to overcome the limitations of classical algorithms and provide a solution adapted to a modern environment, where power efficiency, transmission speed, transparent integration and the quality of advanced image post-processing techniques are central to the user experience.

Traditionally, lossless compression algorithms have been power-hungry, and provided a compression ratio of 2:1 at most. Lossy compression algorithms had been designed mostly for distribution, i.e. for developed images seen by the human eye. When these “visually-lossless” algorithms are used on raw images, they limits the performance of advanced post-processing techniques that rely on recovering weak or hidden signals.

Jetraw achieves the high compression ratios of lossy compression while preserving the advantages raw images have for processing. Minute signals, even when buried in noise, remain recoverable. This is particularly relevant for recent Machine Learning techniques that have been shown to achieve superior performance when applied to raw images (better image quality, speed and memory usage). Any technique that works on raw images will work unmodified on jetraw images.

To achieve its high performance across the above mentioned metrics, Jetraw uses a physical model of the image sensor when used under the specific sensor settings. The calibration parameters are saved into a “.dat” file which then typically contains all the information needed to prepare images from a given camera model (e.g. Leica M11) under all its settings. Once images are prepared, they can be losslessly compressed/decompressed.

Jetraw Platform

Jetraw Core is the implementation of the Jetraw image compression algorithm, and comes as a Command Line Interface, a shared library, a GUI and a VHDL implementation, for deployment to an FPGA.

Contents