机器视觉高斯拟合法自动导星定心系统设计

Design of an Auto-Guiding System with Machine Vision and the Gaussian-Fitting Method

丽江2.4 m光学望远镜自动导星系统升级改造需要计算CCD图像中实际星像中心的位置,在现有导星定心算法的基础上,利用机器视觉Canny边缘检测和椭圆拟合的方法识别星像轮廓,然后对星像直接二维高斯拟合,计算星像中心位置。研究了导星算法中涉及的天光背景参数、星像轮廓识别所需阈值、星像拟合边界参数的整定方法,总结出有效的自动导星定心方法,并开发了基于Linux平台的高速自动导星定心系统软件。软件对星像的高斯拟合结果与IRAF软件高斯拟合结果一致;同一天区相邻时间图像中星像中心偏移量计算结果也表明算法的有效性。

The upgrade and development of the auto-guiding system of the Lijiang 2.4m optical telescope need to incorporate accurate calculations of positions of stars in CCD images. Based on studying the available methods of stellar positioning, this paper gives a prescript of detecting edges of stellar images based on the Canny method with machine vision in conjunction with the method of recognizing stellar profiles with ellipse fitting. We subsequently use the 2-D Gaussian fitting to derive the stellar positions. We have studied the parameter-tuning method to estimate the sky brightness, the threshold, and the boundary coefficients required for fitting. The method is put into an efficient program of auto-guiding. In addition, we have developed a software under the Linux for a high-speed auto-guiding system. Results of our software agree well with those from the 2-D Gaussian fitting provided by the IRAF software. The drifts of the stellar positions calculated with the system from the images in adjacent time steps of the same sly survey further show the validity of the algorithm.