Image retrieval on compressed images: Can we tell the difference?
Juan M. Banda, Rafal A. Angryk, Michael A. Schuh, P. C. H. Martens · 2014
In this work, we discuss the benefits of image compression on FITS image files to perform image retrieval tasks on the enormous NASA Solar Dynamics Observatory (SDO) image repository. With the objective of making solar image files more portable and easy to distribute and archive, we test several lossless compression algorithms as well as lossy compression algorithms in order to determine the rate we can compress standard FITS solar image files and still produce equal or comparable image processing and retrieval results. Our analysis comes from an image processing and retrieval viewpoints since we want to determine if the compression algorithms can reduce storage costs and analysis time. We believe that we might be able to hold huge repositories such as the SDO repository in a considerably smaller amount of disk space and still be able to perform the same image analysis experiments on this reduced and more portable repository.