高能天体物理研究组知识库

首先介绍了几个相关的深度巡天;随后介绍了目前深度巡天中常用的几种恒星-星系自动识别方法,包括双参数空间识别法、神经网络法、图像轮廓拟合法、光谱能量分布拟合法,并结合近期的研究进展对这些方法在处理恒星-星系自动识别问题时的具体适用范围和精度作了系统的介绍和评述;最后,结合有关方法对深度巡天中恒星-星系自动识别问题的研究现状和趋势作了简单总结和展望。

Deep sky surveys provide fundamental data for studies of formation and evolution of stars, galaxies (including the Galaxy), galaxy clusters, large-scale structures of the universe, and supermassive black holes. Currently, optical/near-infrared deep sky surveys are particularly important because of their relatively high depths and resolutions. Stars and galaxies constitute most of the point sources and extended sources in optical/near-infrared deep sky surveys, respectively. Automated classification of numerous detected objects into stars and galaxies is essential for further analyzing data from optical/near-infrared deep sky surveys. Methods for tackling the problem, termed as the problem of automated star-galaxy separation, are based on image-profile or spectral energy distribution (SED) differences between stars and galaxies. Based on differences on the two aspects a set of methods have been proposed and used for automated star-galaxy separation in deep optical/near-infrared sky surveys. These include separations in double-parameter spaces (e.g., magnitude-peak intensity and magnitude-half light radius spaces), classifications with artificial neural networks, image profile fitting and SED-template fitting. In this paper, we first briefly describe some deep sky surveys relevant to research of automated star-galaxy separation. We then list and discuss main existing methods of stargalaxy separation for optical/near-infrared deep sky surveys. We subsequently discuss the current status of applications of these methods to actual deep sky surveys, particularly by paying attention to accuracies and applicabilities of the methods in tackling the star-galaxy separation problem. Finally, we summarize the progress of research of automated stargalaxy separation in deep sky surveys, and give some prospects about future developments for solving the problem.