Snake-based Technique for Automated Coronal Loop Segmentation

Abstract

A new approach to automatically segment the solar coronal loop structures from intensity images of the Sun’s corona is described. The approach is based on the active contour models (snakes) and exploits the Gaussian-like shape of the coronal loop cross-sectional intensity profile to refine the snake’s position. The approach utilizes the principal component analysis to automatically initialize the snake’s position. It then uses a greedy minimization method to attract the snake toward the center of coronal loop structures in each image. Its effectiveness is evaluated through experiments on synthetic and real images.