“…Region growing and thresholding are perhaps the most commonly employed and simplistic of feature detection methods and are well suited to features that appear as a reasonably coherent collection of lighter or darker pixels, such as sunspots (Chapman and Groisman, 1984;Steinegger et al, 1990;Chapman, Cookson, and Hoyt, 1994;Preminger, Walton, and Chapman, 2001;Curto, Blanca, and Martínez, 2008;Zharkov, Zharkova, and Ipson, 2005;Colak and Qahwaji, 2008) or solar filaments (Qu et al, 2005;Aboudarham et al, 2008;Scholl and Habbal, 2008;Shih and Kowalski, 2003;Gao, Wang, and Zhou, 2002). More advanced methods are needed where the features being analysed are not well defined or homogeneous; for example active regions (Shih and Kowalski, 2003;Revathy, Lekshmi, and Nayar, 2005;Delouille et al, 2005). Feature tracking and monitoring algorithms are logical extensions of the detection and classification tasks and also fall into this group.…”