Research On The Extraction Technology Of Coronal Dimming Phenomenon Based On Graph Theory

As one of the most violent and large-scale solar activity, CME (Coronal Mass Ejection) is considered to be the main solar eruptions of affecting the near-earth space environment and space weather. To understand the physical mechanism and the origin of CME, and establish CME forecast system is the hotspot of solar physics as well as the foundation of space weather prediction and model building. As an important physical phenomenon along with CME, coronal dimming is considered to be an important daily surface phenomenon closely related to CME. Therefore, this thesis takes coronal dimming as the research object. With the development of the research, the current research goes deep into the quantitative physical relationship between CME and the coronal dimming, which requires us to accurately locate the dimming region. At the same time, with continuous development of CME observations, the observed data becomes larger and larger, but the extraction techniques of coronal dimming have developed slowly and need to be improved. Therefore, this thesis focuses on the extraction of multiple coronal dimming by using of Graph Cuts method and the gradient, texture, prior knowledge of corona dimming images. It realizes the automatic extraction of corona dimming, and lays the foundation for further study of the relationship between corona dimming and CME.This thesis firstly introduces the main observational approaches and the extraction method’s present research situation of corona dimming. Then it overviews the basics of solar physics and relevant techniques. Finally, it mainly focuses on three aspects of the method for the extraction of coronal dimming:(1) The automatic extraction method of seeds in coronal dimming image. Firstly, this thesis obtains the initial seeds by combining the saliency map of BD and PBD and the prior knowledge of coronal dimming. Then it uses the improved k-means algorithm to cluster the seeds. Finally, the final seeds are obtained successfully by using the empirical threshold. And the whole method does not need manual intervention.(2) The extraction method of corona dimming which combines texture features with gradient features. The method makes use of the multichannel dimension reduction properties of 2D Gabor filter. At the same time, the first extraction result of dimming region is used as an input of the second extraction to obtain the final dimming region. Experimental results show that the proposed method is effective.(3) The automatic extraction system of coronal dimming. One hundred and thirty images of the coronal dimming in 1997-2014 are collected to build the image library and benchmark library of coronal dimming. At the same time, the automatic extraction system of coronal dimming is designed on the basis of the research on extraction algorithm. The designed system realizes the visualization from the image reading to analysis of dimming region.