Notes On Gaussian Fitting To A Stellar Image

At present, computers and digital image processing techniques have been widely applied to biomedical, remote sensing, military, astronomical and many other disciplines and fields. The centering algorithms in digital image techniques can be used to determine the image center of a target object. Among the measurement of celestial bodies, the center of a stellar image is of fundamental and physical information. In order to achieve even greater accuracies, the image processing method still plays an important role.In this paper, we applied the distribution of chi-square Gaussian fitting to measuring stellar center, and made comparison with the traditional two-dimensional Gaussian fitting results. In practice, some real CCD images from actual observations showed that the precision of the measured positions with the two algorithms were almost equal. On the other hand, we explored the precise positioning technique when a part of the stellar distribution was not included in a scanning box. The results showed that, if the full width at half maximum (FWHM) derived by other stars in the same field was adopted and only the remaining four parameters (except for FWHM) were fitted to a stellar image by 2-dimension Gaussian, positional errors were found to be significantly reduced by comparison with former results from Stone. Specifically, along each direction of x- or y- coordinate, when a stellar center was shifted at 1FWHM away from a box boundary, the positional error would be less than 1 mas.In this paper, based on Visual C++/Windows integrated development environment, our developed program can display a FITS format image, and execute a Chi-square Gaussian fitting and a traditional two-dimensional Gaussian fitting to a distribution of a stellar image to derive its centers.