Denoising observational data

Abstract

Reducing noise caused by the instrumentation in observational data is a crucial step in data post-processing. A method is searched for that conserves most of the instrumental resolution and introduces as few methodical artefacts as possible. With such a method integrated in an observation sites software tool-chain, the resources spent for the generation of observational data will more likely find their way into resulting scientific publications; otherwise, for data post-processing often methods are used, which just smear out the noise, introduce artefacts, or decrease the provided resolution in space or time. A short review of different techniques is given here, and a non-local averaging method is applied to Hinode magnetograms and G-band data. The presented method fits the needs for various kinds of observational data.