A detection method of coronal mass ejection based on improved u-net network

Abstract

The invention discloses a coronal mass ejection detection method based on an improved U-Net network, comprising the following steps: step 1, performing image preprocessing on the LASCO C2 image downloaded from the SOHO homepage, and the preprocessing includes image grayscale , image rotation and flipping operations; step 2, convert the LASCO C2 image obtained through step 1 into polar coordinates; step 3, make a coronal image data set, and divide the data set into a training set and a test set; step 4, Improve the original U‑Net network to adapt it to the CME detection task; step 5, use the improved U‑Net network for training, fine-tune the network parameters, and finally obtain the detection results. Realize the automatic detection of the coronal mass ejection phenomenon, reduce the difficulty of manual recording, and speed up the automatic detection speed.

Description

technical field [0001] The invention relates to the field of astronomical image target detection, in particular to a coronal mass ejection detection method based on an improved U-Net network. Background technique [0002] Coronal Mass Ejections (CMEs) are a frequent eruption phenomenon in the solar atmosphere. The eruption time and frequency vary with the frequency of solar activity, which is manifested in the coronal structure within the time interval of several minutes to several hours. Significant changes accompanied by observable mass ejections typically appear as a bright, intricately textured enhancement structure, often accompanied by a dark region of insufficient brightness. There is no clear definition of CMEs, and earlier definitions were visible changes in the structure of the corona, including the appearance and outward movement of a new, discrete, bright white light feature in the coronal field of view. CMEs are composed of large structures containing plasma an...

AI Technical Summary

Problems solved by technology

[0006] The above methods have more or less certain problems. For example, the methods based on grayscale features, such as CACTus and SEEDS, all use brightness enhancement to highlight the suspected CMEs area in the coronal image, and enhance the target area of ​​CMEs. The intensity of the CMEs area, and the detection results will be greatly affected by the detection rules and threshold selections set by each

Based on the optical flow algorithm, the velocity vector of each pixel can be estimated from the continuous coronal image sequence to form an image motion field, which can determine the obvious moving target, but the detection and tracking effect for weaker CMEs is poor, and it is impossible to distinguish similar coronal mass ejections. sun structure

Images