基于实时平场的NVST观测数据干涉条纹消除

NVST observations, both monochromatic image and optical spectrum, are always troubled by interference fringes. The project is aimed at solving the problem that interference fringes underlying NVST observation data cannot be effectively eliminated by using the standard flat-fielding method. Considering the interference fringes as a slowly time-varying flat-fielding model, this project investigates the interference fringes elimination method from the data processing perspective. This project aims to establish the real-time flat-fielding theory and method when the interference fringes underlying the observations slowly change, and to explore the real-time flat-fielding method by combining maximum correntropy criterion with online observations; to analyze and extract the interference fringe feature from the observation data by using time-frequency analysis tools, and to create time-varying mathematical model of the interference fringes, and to realize the automatic elimination of interference fringes underlying observations. In practice, we verify the proposed new technique with the NVST observations, and analyze the effectiveness of the restored image and promote the real-time flat-fielding technique to practical application. This project is expected to explore the automatic detection and removal method of the interference fringes underlying the NVST observations degraded by the hardware, develop the automatic elimination technique of time-varying interference fringes, explore the real-time flat-fielding technique, achieve some breakthroughs on the data restoration techniques of the NVST observations, and has important significance of improving the quality of astronomical observation product.

NVST观测数据（无论是单色像还是光谱）一直受到了干涉条纹问题的困扰。针对标准平场算法不能有效消除NVST观测数据中干涉条纹这一问题，把干涉条纹看成是一个缓慢时变平场模型，本项目从数据处理角度，1)研究缓变条件下的实时平场原理和方法，利用实时观测数据，结合相关熵最大准则，探索实时平场方法；2)利用时频工具分析和提取干涉条纹的特征，建立时变干涉条纹的数学模型，实现干涉条纹的自动消除。实践中以NVST观测数据中干涉条纹消除为例，分析还原数据的有效性，以验证这一新技术，将实时平场技术与方法推向实用化。本项研究旨在从实际科研需求出发，针对与硬件关联的数据质量问题，研究时变干涉条纹自动消除，探索实时平场新技术，在NVST观测数据还原技术和方法上有所突破，提高天文数据产品质量，具有重要意义。

中国科学院云南天文台抚仙湖观测站NVST观测数据一直受到了干涉条纹问题的困扰。针对标准平场算法不能有效消除NVST观测数据中干涉条纹这一问题，本项目把NVST观测数据中的干涉条纹看成是一个缓慢时变平场模型，从图像数据处理的角度，建立NVST干涉条纹平场模型，研究基于实时平场的NVST干涉条纹自动去除方法，结合2013年到2018年的部分NVST太阳色球观测数据的处理实践，将基于实时平场的干涉条纹去除技术与方法推向实用化。项目成果在NVST观测数据条纹识别和去除的技术和方法上有所突破，提高了NVST天文数据产品的质量。1、分析包含干涉条纹的NVST数据中条纹缓慢变换的空时特性，提出了一套实时条纹平场的提取方法；2、利用时频分析/小波变换等工具，研究干涉条纹时变频谱的变化规律，提出了分别针对原始数据以及高分辨数据的干涉条纹的自动去除算法；3、针对NVST观测数据，采用深度学习的方法，提出了NVST干涉条纹的自动识别方法，自动区分有无条纹的原始数据。项目成果在Solar Physics、Publications of the Astronomical Society of the Pacific、Astronomy and Computing、Neurocomputing等国际期刊发表论文9篇，《天文学报》论文2篇，培养硕士9名。