DOI: 10.1016/J.JVCIR.2015.09.015
关键词: Region of interest 、 Mathematical morphology 、 Coronal hole 、 Image segmentation 、 Segmentation 、 Scale-space segmentation 、 Artificial intelligence 、 Computer science 、 Coronal plane 、 Solar physics 、 Computer vision
摘要: After the seed is indicated, segmentation process progresses automatically.A new algorithm for merging regions into output mask of coronal hole.The average time a single hole 342ms.Based on experiments completed, Dice's index averages 89%.High accuracy results holes irregular shape. Region interest in solar images subject frequent research physics. This study outlines watershed by immersion to identify areas acquired using Extreme UV Imaging Telescope (EIT). Solutions presented here produce highly accurate shape, and what more, they do so representing varied activity, recorded different years months. In addition, solutions make all methods used operate very quickly. These include: preprocessing step before segmentation, itself, also postprocessing after segmentation. The mean duration entire amounts 342ms hole, without parallel implementation used. were carried out computer with an Intel Core i7 CPU @ 2GHz 4GB RAM. point identified inside runs automatically.