Preparation for CSST:Star-galaxy Classification using a Rotationally Invariant Supervised Machine Learning Method

作     者：Shiliang Zhang Guanwen Fang Jie Song Ran Li Yizhou Gu Zesen Lin Chichun Zhou Yao Dai Xu Kong Shiliang Zhang;Guanwen Fang;Jie Song;Ran Li;Yizhou Gu;Zesen Lin;Chichun Zhou;Yao Dai;Xu Kong

作者机构：Institute of Astronomy and AstrophysicsAnqing Normal UniversityAnqing 246133China Deep Space Exploration Laboratory/Department of AstronomyUniversity of Science and Technology of ChinaHefei 230026China School of Astronomy and Space ScienceUniversity of Science and Technology of ChinaHefei 230026China National Astronomical ObservatoriesChinese Academy of SciencesBeijing 100101China Tsung-Dao Lee Institute and Key Laboratory for Particle PhysicsAstrophysics and CosmologyMinistry of EducationShanghai Jiao Tong UniversityShanghai 200240China Department of PhysicsThe Chinese University of Hong KongShatinN.T.Hong KongS.A.R.China School of EngineeringDali UniversityDali 671003China 

出 版 物：《Research in Astronomy and Astrophysics》 (天文和天体物理学研究（英文版）)

年 卷 期：2024年第24卷第9期

页      面：136-146页

核心收录：

学科分类：07[理学] 070401[理学-天体物理] 0704[理学-天文学] 

基　　金：supported by the Strategic Priority Research Program of Chinese Academy of Sciences(grant No.XDB41000000) the National Natural Science Foundation of China(NSFC,Grant Nos.12233008 and 11973038) the China Manned Space Project(No.CMS-CSST-2021-A07) the Cyrus Chun Ying Tang Foundations the support from Hong Kong Innovation and Technology Fund through the Research Talent Hub program(GSP028) 

主　　题：methods:data analysis techniques:image processing stars:imaging 

摘      要：Most existing star-galaxy classifiers depend on the reduced information from catalogs,necessitating careful data processing and feature *** this study,we employ a supervised machine learning method(GoogLeNet)to automatically classify stars and galaxies in the COSMOS *** traditional machine learning methods,we introduce several preprocessing techniques,including noise reduction and the unwrapping of denoised images in polar coordinates,applied to our carefully selected samples of stars and *** dividing the selected samples into training and validation sets in an 8:2 ratio,we evaluate the performance of the GoogLeNet model in distinguishing between stars and *** results indicate that the GoogLeNet model is highly effective,achieving accuracies of 99.6% and 99.9% for stars and galaxies,***,by comparing the results with and without preprocessing,we find that preprocessing can significantly improve classification accuracy(by approximately 2.0% to 6.0%)when the images are *** preparation for the future launch of the China Space Station Telescope(CSST),we also evaluate the performance of the GoogLeNet model on the CSST simulation *** results demonstrate a high level of accuracy(approximately 99.8%),indicating that this model can be effectively utilized for future observations with the CSST.