ePSF拟合法与Gaussian拟合法的比较

Comparison of Fitting ePSF and Fitting Gaussian-Functions

为了解决哈勃空间望远镜的图像欠采样问题,美国学者Anderson和King提出了精确测量星像位置和通量的有效点扩散函数(ePSF—effective point-spread function)拟合法。然而,他们却不加比较地将其应用于地面CCD图像中星像的位置高精度测量。因此,我们试图将ePSF拟合法与经典的高斯函数(Gaussian)拟合法作比较研究。调用CFITSIO库生成模拟的背景图像,并应用不同参数条件下的星像轮廓模型产生非欠采样的星像。最后,分别采用ePSF拟合法与Gaussian拟合法对这些星像进行拟合,并对它们的拟合精度进行比较。实验结果表明,在非欠采样的图像中这两种算法对星像位置的测量几乎是等精度的。

In order to solve the problem of point spread function（PSF） undersampling for the images from the Hubble Space Telescope （HST）, Anderson and King proposed an effective PSF model（ePSF）with which high-precision stellar positions and fluxes can be derived through fitting. However, they applied it to ground-based images without comparing with any classical methods. We thus compare fitting ePSF to the classical fitting with Gaussian functions. We used the CFITSIO Library to simulate background images, and superimpose simulated stellar images （not undersampled ） corresponding to stellar profiles of different parameters on the background images. Finally, we fit ePSF and Gaussian functions to these stellar images separately, and make the comparison for their fitting precisions. The preliminary experiments show that the precisions of the positions measured with the two algorithms are nearly identical.