Automated recognition of sunspots on the SOHO/MDI white light solar images

Abstract

A new technique is presented for automatic identification of sunspots on the full disk solar images allowing robust detection of sunspots on images obtained from space and ground observations, which may be distorted by weather conditions and instrumental artefacts. The technique applies image cleaning procedures for elimination of limb darkening, intensity noise and noncircular image shape. Sobel edge-detection is applied to find sunspot candidates. Morphological operations are then used to filter out noise and define a local neighbourhood background via thresholding, with threshold levels defined as a function of the quiet sun intensity and local statistical properties. The technique was tested on one year (2002) of full disk SOHO/MDI white light (WL) images. The detection results are in very good agreement with the Meudon manual synoptic maps as well as with the Locarno Observatory Sunspot manual drawings. The detection results from WL observations are crossreferenced with the SOHO/MDI magnetogram data for verification purposes.