An adaptive weighted centroid localization algorithm is proposed to solve the problems caused compositely by the high magnitude stars, the image motion and the aberration of optical system in high accuracy star localization. A mathematical model based on the maximum likelihood estimation is derived and built to perform numerical simulations. The results show that the localization accuracy of the algorithm proposed is increased by over 50% compared with that of the traditional star localization algorithms such as centroid method, weighted centroid method and Gaussian fitting method. The proposed algorithm has good convergence rate, and the centroid location error is relatively stable after five iterations, which can meet the real-time requirement in practical applications.<>