摘要
星图识别提取是自动天文测量数据处理的关键环节.针对视频测量机器人拍摄的小视场星图具有高噪声的特性,提出了利用基于合理阈值分割的连通算法来处理此类星图;分析了4种不同星图的图形特征,对比了常见的星点提取算法对真实星图的识别效果;在定性定量分析的基础上,通过验证,得到了阈值分割下的连通算法要优于边缘检测和聚类算法的结论,并获得理想的星点提取效果.真实星空半仿真星图的室内试验表明:利用该算法可以准确可靠地计算恒星质心坐标,其水平和垂直方向均方根误差(RMSE)分别为0.025和0.021像素,可满足高精度的天文测量需求.
Star map recognition and extraction is key to automatic astrometric data processing.In view of the high noise characteristics of small field star images captured by video measurement robot,a connectivity algorithm based on reasonable threshold segmentation is proposed to process such star images.The graphic features of four different star images were analyzed,recognition effect of common star point extraction algorithms on real star images were compared.Qualitative and quantitative analysis verified that connected algorithm under threshold segmentation could obtain perfect and ideal star point extraction,better than edge detection algorithm and clustering algorithm.Indoor semi-simulated star images based on real star sky confirmed that this algorithm was accurate and reliable.The root mean square errors in horizontal and vertical directions were 0.025 and 0.021 pixels respectively,meeting needs of high-precision astronomical measurement.
作者
张超
时春霖
吴建霖
于广瑞
陈少杰
焦博
陈长远
汤进九
ZHANG Chao;SHI Chunlin;WU Jianlin;YU Guangrui;CHEN Shaojie;JIAO Bo;CHEN Changyuan;TANG Jinjiu(Geospatial Information Academy,Information Engineering University,450001,Zhengzhou,Henan,China;Troop 61206,100042,Beijing,China;Troop 63620,732750,Jiuquan,Gansu,China;National Time Service Centre,Chinese Academy of Sciences,710600,Xi’an,Shaanxi,China;University of Chinese Academy of Sciences,100049,Beijing,China;Troop 61365,300100,Tianjin,China)
出处
《北京师范大学学报（自然科学版）》
CAS
CSCD
北大核心
2022年第2期193-202,共10页
Journal of Beijing Normal University(Natural Science)
基金
国家自然科学基金资助项目(41804034,41804031,42074013)。
关键词
自动天文测量
小视场星图
星点提取
阈值分割
边缘检测
聚类分析
automatic astronomical survey
star map with small field of view
star point extraction
threshold segmentation
edge detection
cluster analysis
