A developed method for filament automatic detection and statistical studies

Hao Qi, hq_turtle@163.com, Nanjing University, China
Fang Cheng, Nanjing University
Chen P.F., Nanjing University

Abstract
We developed a method to automatically detect and tract solar filaments from H$\alpha$ full disk solar images that are obtained by Mauna Loa Solar Observatory (MLSO). The program is able to not only recognize filaments, determine their features such as position, area, spine and other relevant parameters, but also trace daily evolution of filaments. The program consists of three parts: first, a pretreatment is applied to correct the original images; second, using Canny edge detection method detects filaments ; third, recognizing filaments features through the morphological operators. The program is proven to be accurate and efficient. In addition, images obtained by MLSO from 1998 to 2010 have been tested and some statistical properties of filaments are also presented.