Arch-shaped coronal loops that are isolated from the background are typically acquired manually from massive online image databases to be used in solar coronal research. The manual search for special coronal loops is not only subject to human mistakes but is also time consuming and tedious. In this study, we propose a completely automated image retrieval system that identifies coronal-loop regions located outside of the solar disk from 17.1 nm EIT images. To achieve this aim, we first apply image preprocessing techniques to bring out loop structures from their background and to reduce the effect of undesired patterns. Then we extract principal contours from the solar image regions. The geometrical attributes of the extracted principal contours reveal loop existence in a given region. Our completely automated decision-making procedure gives promising results in separating the regions with loops from the regions without loops. Based on our loop-detection procedure, we have developed an automated image-retrieval tool that is capable of retrieving images containing loops from a collection of solar images.
Authors: Nurcan Durak , Olfa Nasraoui, Joan Schmelz
Publication Status: Accepted to Solar Physics
Last Modified: 2010-04-30 07:49