A Threshold and Region Growing Combined Method for Filament Disappearance Area Detection in Solar Images

This paper presents one efficient automatic detection method for filament disappearance. The method was applied to the Big Bear Solar Observatory (BBSO) fulldisk Hα images. The initial step is to detect the filament pixels by the threshold method. Then, using a regiongrowing method, we merge the adjacent detected pixels to obtain all filament areas. Based on these two steps, an automatic filament detection system is developed. This system is proven to be accurate and fast. The detection results allow one to detect filament disappearances in real time, to get its location, size, time and to generate a warning message in a timely fashion. The detected filaments and filament disappearances in 1999 are analyzed in this work. Some statistical results are presented to prove that the system works reasonably well.