Lossless compression using joint predictor for astronomical images

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review


Downloading astronomical images through Internet is a slow operation due to their huge size. Although several lossless image coding standards that have good performance have been developed in the past years, none of them are specifically designed for astronomical data. Motivated by this, this paper proposes a lossless coding scheme for astronomical image compressions. We design a joint predictor which combines the interpolation predictor and partial MMSE predictor. Such strategy benefits from its high compression ratio and low computation complexity. Moreover, the scalable and embedding functions can be further supported. The interpolation predictor is realized by upsampling the downsampled input image using bi-cubic interpolation, while the partial minimum mean square error (MMSE) predictor predicts the background and foreground (i.e., stars) separately. Finally, we design a simplified Tier-1 coder from JPEG2000 for entropy coding. Our experimental results show that the proposed encoder can achieve higher compression ratio than JPEG2000 and JPEG-LS.

Original languageEnglish
Title of host publicationAdvances in Visual Computing - 5th International Symposium, ISVC 2009, Proceedings
Number of pages9
EditionPART 2
StatePublished - 2009
Event5th International Symposium on Advances in Visual Computing, ISVC 2009 - Las Vegas, NV, United States
Duration: 30 Nov 20092 Dec 2009

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
NumberPART 2
Volume5876 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349


Conference5th International Symposium on Advances in Visual Computing, ISVC 2009
Country/TerritoryUnited States
CityLas Vegas, NV


Dive into the research topics of 'Lossless compression using joint predictor for astronomical images'. Together they form a unique fingerprint.

Cite this