Abstract: An on-orbit calibration method of star sensor based on Kalman filter is proposed in this article. Considering the distortion feature of most optical camera lens, internal and external parameters of star sensor are calibrated separately using Kalman filter with first-order radial distortion. Taking the image coordinates and the corresponding celestial coordinates of observed star points as the filter inputs and taking the estimated value of the internal and external star sensor camera parameters as the filter outputs, the optimized value of the internal and external parameters are obtained with Kalman filter algorithm, while state equation and observation equation for Kalman filter are established based on radial alignment constraint (RAC) which is widely used in camera calibration of computer vision. The simulation results indicate that this method which is independent on external attitudes can eliminate the coupling between external parameters and internal parameters effectively, and it also satisfies the optimal qualification demanded by Kalman filter that both state equation and observation equation are linear. When the standard deviation of star positional noise is set to 0.05 pixels, the internal and external parameters are correctly obtained and the residual errors can reach 0.044 pixel on average in x direction and 0.049 pixel in y direction respectively.

Key words: star sensor, Kalman filters, on-orbit calibration, radial distortion model, radial alignment constraint