Astronomical Image Compression Techniques Based on ACC and KLT Coder

Schindler, Jaromír; Páta, Petr; Klíma, Miloš; Fliegel, Karel

doi:10.14311/1336

Cited by 1 publication

(1 citation statement)

References 4 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A second contribution is the gathering and publication of a new test data set comprising real data from multiple telescopes. This corpus contains a larger amount of samples depicting a wider variety of astronomical objects than previously described corpora (Schindler et al 2011;Kitaeff et al 2015;Pata & Schindler 2015), and is intended to be representative of modern observatories. Only the work by Pence et al (2009) reports results for a larger data set, including sky and calibration images.…”

Section: Introductionmentioning

confidence: 99%

Efficient Lossless Compression of Integer Astronomical Data

Maireles-González,

Bartrina-Rapesta,

Hernández-Cabronero

et al. 2023

PASP

View full text Add to dashboard Cite

Each new generation of telescope produces increasingly larger astronomical data volumes, which are expected to reach the order of exabytes in the next decade. Effective and fast data compression methods are paramount to help the scientific community contain storage costs and improve transmission times. Astronomical data differs significantly from natural and Earth-observation images, asking for specifically tailored compression approaches. This paper presents a novel lossless compression technique that employs the discrete Haar wavelet transform within the JPEG 2000 standard. Its performance is compared to that of a comprehensive selection of compressors, including fpack, the most common technique in astronomical observatories, as well as other algorithms highly competitive for other types of data. Experiments are performed on a large data set of 16 bit integer images, produced by telescopes around the world and representative of a wide variety of astronomical scenarios. The proposed technique has two modes. The first mode outperforms all the other tested techniques in terms of compression performance. It surpasses the most competitive configuration of fpack by, respectively, 5.3% (about 0.3 bits per sample), having also 4.5% lower compression and decompression times. The second mode is the fastest among all tested techniques. Its compression and decompression times are 2.5 and 3.5 times faster than the fastest configuration of fpack, while also yielding a 2.4% better compression performance (0.15 bits per sample).

show abstract