Image data compression with FAPEC
Document typeMaster thesis
Rights accessOpen Access
Satellite data compression is very important for scientific missions, which often send large amounts of data during operations. Modern examples are Euclid, Solar Orbiter and the several Earth observation missions. However, for each mission and type of data the optimum solution can be different. Here we focus on compression of simulated images following the design specifications of the Euclid and Solar Orbiter missions, and on real images from NOAA and Eumetsat Earth Observation missions. We aim to achieve the best solution for the compression of these images, mainly regarding compression ratio, but also paying attention to the compression time - that is, trying to minimize the onboard processing requirements. Data compression research at the Institute for Space Studies of Catalonia (IEEC) has led to new algorithms that in some cases perform better than the consolidated standards of the CCSDS (Consultative Committee for Space Data Systems). The so-called Fully Adaptive Prediction Error Coder (FAPEC) is a new entropy-coding algorithm for data compression that provides an excellent coding efficiency even when large fractions of outliers are present in the data. It competes with the CCSDS 121.0 recommendation. On the other hand, the CCSDS 122.0 recommendation for image compression, based on the Discrete Wavelet Transform (DWT), can be used to produce both lossy and lossless compression. Previous studies have combined FAPEC with the 122.0 standard by compressing non-scaled DC and AC coefficients from the DWT stage, allowing FAPEC to achieve better compression ratios than the standard bit-plane encoder. The result, called DWTFAPEC, was able to work with FITS and raw images, both lossless and lossy. In this work we modify DWTFAPEC to differentially code the DC coefficients, further improving the performance - up to 10% in the lossy case. DWTFAPEC has been tested on a variety of realistic images (from Euclid, Solar Orbiter and Earth observation missions) to evaluate its performance and to demonstrate its eventual applicability to any space mission. For each of the cases studied here we show which is the optimum data compression solution among FAPEC, DWTFAPEC, CCSDS 121.0 and 122.0, both for lossless and lossy (where applicable), and for different options of the compressors. In all cases we consider both compression ratios and computing times. In some cases, due to the lack of capability to work with colour images we partitioned the images in sub-streams - one for each of the colour bands. This approach makes possible to use DWTFAPEC and FAPEC for colour images.