For more than 15 years, DIGINOVE designs and develops the heart of the image processing software for the ground segments of many french civil and military observation satellites.
The performance and the flexibility of its image processing software engine allow our products to process very large images (hundreds of gigabytes, as most of the recent satellites produce) as well as more traditional documents, but also images with complex components (panchromatic, multispectral, hyperspectral or same radar) high accuracy (more than 16 bits per pixel) and besides, manage the associated geolocation with precision.
After having tested the technology with a “proof-of-concept” developed for SpotImage in the 1990s, we developed our first production software in the early 2000s for the DGA, and then another one for the DRM. We also took part in the European EOLES projects for rapid dissemination of satellite image in the event of a natural disaster, and in the IMPAST project to help detection of illegal fishing boats.
In recent years, we have continued to expand our expertise in providing the image processing and image diffusion modules for the Pharos project (Portail Hôte d’Accès au Renseignement d’Origine Spatiale) in 2010. In 2014, we started to work to his successor the SSU CSO (Segment Sol Utilisateur de la Composante Spatiale Optique) which is an extension to the latest technologies of space imaging.
DIGINOVE and the user ground segment modules
As opposed to the space segment, the ground segment is a chain of the processing modules located on the ground. They prepare and disseminate the satellite products to their users.
In this context, we have made:
- an ingestion module, which allows to convert the raw products to a unified format: the XXL format
- a remote visualisation module especially adapted to low bandwidth networks: using some of the most advanced features of the JPEG2000 compression algorithm, the XXL format can be exploited on is a lossless format, that means that the quality of thproduct can be regenerated identicaly but it can also be comfortably visualized from remote locations
- several more functional modules based on the visualisation module (validation of an acquisition, production request, delivery request, etc.)
- a transfer module that transfer at any remote station a lossless or degraded, and thus less voluminous, version of the product depending on the user choice
- a module that allows from a transfered product to regenerate this product in his original format
The XXL format
This public format, developed by DIGINOVE, is based on a pyramid of tiles using JPEG2000 compression algorithm. It has the capability to contain not only the “pixels” of the images but also all associated metadata, including of course the geolocation of the product.
It can thus be used to store products from all sources in a unified lossless format, whether it’s optical products, radar, IR or hyper-spectraux.
The pyramidal structure is designed to make the remote access performance independent of the size of the image. You can navigate in a product multi-gigaoctets as comfortably as if it was only a few megabytes, even through very low bandwidth networks.
Using the progressive nature of JPEG2000, it also allows to transfer only a degraded version of the product to make the transfer much quicker:
- it can degrade the resolution in most areas of the product and keep the full resolution on the sensitive parts
- it can degrade the quality in all or part of the product and keep lossless information on special points of interest
- or combine any of the two modes
The gain in time and information volume may be important as tests made with professional interprets showed that a product may be reduced to one fifth of its original volume without that a human eye, even very exercised, see any difference. Of course, if the analysis is made using very sensitive image processing software that need the whole product to be left untouched, the user may transfer a lossless version of the product and thus keep intact all the original information.