基于日球物理学事件知识库的太阳活动识别

发布时间：2018-03-24 09:02

本文选题：太阳活动识别　切入点：图像分割　出处：《昆明理工大学》2017年硕士论文

【摘要】：基于内容的图像检索技术发展至今,已经在天文图像处理领域得到了广泛应用。随着多波段太阳观测技术的迅猛发展,所采集的太阳图像分辨率越来越高,极大地促进了太阳活动研究的进展。相应地,海量天文数据的激增,也给天文图像的数据存储带来了极大的挑战。而如何从全日面图像中自动检测和识别出有效的太阳活动,是天文图像处理领域的另一个难题。本文基于以上两个问题展开研究,包含以下三个方面:第一,本文提出了一种基于方形网格结构的太阳活动目标检测方法(GBTD)。此方法将太阳图像划分成等大小的方形网格结构,基于多阈值选取策略和GBTD策略分离出目标区域和背景区域。通过对6种太阳活动,共2172个太阳活动区域进行分割实验,结果表明,该方法在切割准确度及时间开销方面取得满意的结果,对图像噪声具有良好的抗干扰性。GBTD方法为多种类型的太阳活动的研究提供一种通用的图像分割方法,也为解决海量天文数据存储的难题提供了一种可行办法。第二,对太阳图像特征参数相关性的研究,得到了每种太阳活动的最佳特征参数组合。对不同太阳活动区域提取特定组合的特征,可以为基于内容的图像检索系统(CBIRS)建立精简的图像特征集提供了一种可行办法。第三,得益于美国太阳动力学天文台(SDO)的日球物理学事件知识库(HEK)所提供的实时太阳观测数据,本文提出了一种基于日球物理学事件知识库的太阳活动识别方法。此方法获取6种太阳活动的信息(发生时间、位置、区域面积),建立对应时间的全日面图像的尺度变换模型。结合位置与区域面积信息,对不同太阳活动进行梯度阈值分割,边界识别方法被用来定位和识别太阳活动的区域。本文方法实现了对太阳活动的精确定位和有效识别,为后续工作的开展提供了便利。
[Abstract]:Content-based image retrieval technology has been widely used in the field of astronomical image processing. With the rapid development of multi-band solar observation technology, the resolution of the collected solar image is getting higher and higher. It has greatly promoted the research progress of solar activity. Accordingly, the explosion of massive astronomical data has also posed a great challenge to the data storage of astronomical images. And how to automatically detect and recognize effective solar activity from all heliospheric images, It is another difficult problem in the field of astronomical image processing. Based on the above two problems, this thesis includes the following three aspects: first, In this paper, a method of detecting solar moving objects based on square grid structure is presented. This method divides the solar image into square grid structures of equal size. The target region and background area are separated based on multi-threshold selection strategy and GBTD strategy. 2172 solar active regions are segmented by six solar activities, and the results show that, The method obtained satisfactory results in terms of cutting accuracy and time cost. The method has good anti-interference to image noise. GBTD method provides a general image segmentation method for the study of various types of solar activity. It also provides a feasible way to solve the problem of storing massive astronomical data. Secondly, the correlation of the characteristic parameters of solar images is studied. The optimal combination of characteristic parameters for each solar activity is obtained. Extracting the features of specific combinations for different solar active regions provides a feasible method for building a simplified image feature set for content-based image retrieval system (CBIRS). Third, Benefiting from the real-time solar observation data provided by the knowledge Base on Solar Sphere Physics events of the United States Solar Dynamics Observatory (SDO), In this paper, a method of solar activity recognition based on the knowledge base of heliospheric physics events is proposed. In this paper, the scale transformation model of the whole heliospheric image corresponding to the time is established, and the gradient threshold segmentation of different solar activity is carried out by combining the information of position and area. The boundary recognition method is used to locate and identify the region of solar activity. In this paper, the accurate location and effective identification of solar activity are realized, which provides convenience for the subsequent work.
【学位授予单位】：昆明理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.41

【参考文献】