Compression Techniques for 4D STEM data in Electron Microscopy

Abstract

This paper explores compression techniques for 4D scanning transmission electron microscopy (STEM) data, aiming to find an optimal, use-case-independent compression approach. Standard image compression methods, including JPEG, PNG, and JPEG2000, are assessed and compared due to their ability to preserve the structural characteristics of the data. Additionally, a method based on Slepian two-dimensional sequences is proposed. It is conceptually similar to JPEG. The methods are evaluated based on compression ratio and data fidelity to identify an optimal format that maximizes data memory demands while minimizing information loss.