Abstract: To extract the star position and star magnitude, a new algorithm was developed. The algorithm regards the star imaging pattern as Gauss point spread function model. The coefficient of Gauss curved face function was getted using the combination of linearity interpolation method and least squares method. From the Gauss curved face function, the sub-pixel position of star and its star magnitude were obtained. Theoretical research indicates that the extracted precision using Gauss curved method is higher than using traditional centroid method. The calculation of the direct Gauss curved interpolation is very complex, for reducing calculation, the x, y direction nonlinear interpolation method was used in imaging point adjacent, and the different coefficients of Gauss curved was getted individually. The simulative calculation results show that the locating precision less than 1/20 pixel and error of star magnitude less than 5% when the signal-to-noise ratio is 0.05.