Abstract
In this paper, a new adopted unsupervised segmentation technique is presented. This technique extracts both sunspots and sunspot groups from the solar disk as whole group that eases the automated sunspot group classification. MacIntosh Classification is used as standard to determine which group belongs to which class. Sunspot data are extracted from daily white light (WL) images of a solar disk captured by the SOHO/MDI satellite. New unsupervised segmentation algorithm is applied to extract sunspot groups from the full disk image with use of mathematical morphology in order to improve the quality and segregate the sunspot groups. The results were compared with the standard data from EGSO, SFC catalogues and with Mt.Wilson Report. The obtained results reveal a good accuracy of group classification, which is very promising in comparison with the manual classification, catalogues.