Evolved transforms for improved reconstruction of lossy-compressed NASA images

Abstract

This paper describes a methodology for evolving image reconstruction transforms consisting of an arbitrarily large, user-selected number of wavelet and scaling numbers. Given images previously subjected to lossy compression using NASA's wavelet-based ICER compressor, these novel transforms are capable of reconstructing those images with less error than ICER's own reconstruction scheme. This advance has the potential to enhance the science value of all images subjected to lossy compression.

DOI: 10.1145/2598394.2602276

Extracted Key Phrases

Cite this paper

@inproceedings{Moore2014EvolvedTF, title={Evolved transforms for improved reconstruction of lossy-compressed NASA images}, author={Frank W. Moore and Brendan J. Babb}, booktitle={GECCO}, year={2014} }