Efficiently storing and accessing high-resolution solar images using light weight compression techniques

Abstract

This thesis is focused on implementing lightweight compression techniques for efficiently storing and accessing high-resolution solar images obtained from Solar Dynamics Observatory. The solar images utilized in this thesis have two-dimensional image data. The value distribution across multiple images is investigated. Images of a different wavelength are likewise subjected to this examination. Dictionary encoding is used as the basic compression layer to compress solar images from AIA for all ten solar wavelengths. Encoding schemes such as Sparse encoding, Run-length Encoding and Coding techniques such as Elias Gamma, Elias Delta and Fibonacci coding are applied as the second layer of compression. A collection of images are stacked to form a multidimensional data cube which is cut into n slices. The compression is applied individually on each of these slices. The evaluation is performed separately on images of various wavelengths, and the compression and decompression outcomes are evaluated independently. The algorithm performance on compressing the images is measured with compression ratio. Various Decompression operations, such as extracting a single slice of a specific image from the cube or extracting a range of slices of multiple images etc., are implemented, and performance of decompression is measured in time.