Fast Segmentation of Solar Extreme Ultraviolet Images

Abstract

A segmentation scheme for identifying large-scale structures (coronal holes, active regions, etc.) in solar extreme ultraviolet images, is presented. Unlike standard approaches, both the image intensity and the relative contribution of different wavelengths are used. Spectral information is important for compensating luminosity changes. The approach is illustrated with images taken in the extreme ultraviolet by the EIT telescope onboard SOHO. This supervised segmentation scheme, which incorporates a Bayesian classifier, is computationally simple, and can easily be used to track in near-real time structures, such as coronal holes.