Gyroscope-Aided Star Localization Algorithm With Adaptive Iterative Richardson–Lucy Algorithm and Interpolation Registration Effective Point Spread Function Construction

To address the degradation of localization accuracy in star trackers under dynamic conditions, this article proposes a gyro-aided localization method integrating trailing star-spot reconstruction with effective point spread function (ePSF) construction. First, motion blur kernels are parameterized using the gyroscope's angular velocity and exposure time, followed by an adaptive iterative Richardson–Lucy (RL) algorithm to restore trailing star spots. A residual minimization-based iteration termination criterion is established to achieve a high signal-to-noise ratio (SNR) star-spot restoration. Second, we estimate the relative displacements of stars across multiple frames using the gyroscope's angular velocity measurements, and then perform subpixel-level nonuniform interpolation registration. By integrating the registered star samples, we reconstruct a high SNR ePSF model. Finally, subpixel localization is implemented using the refined ePSF. Validated by numerical simulations, laboratory experiments, and real night sky observations, the proposed method reduces localization error by approximately 23% compared to the traditional gyro-free ePSF construction method (0.122 pixel) under 3°/s dynamic conditions, achieving a localization accuracy of 0.094 pixel. This demonstrates that by parameterizing the blur kernel with gyroscope information and constraining subpixel displacements, the present approach eliminates both kernel estimation and registration errors inherent in conventional methods, thereby providing theoretical and practical support for approaching the static precision limit (PL) in dynamic situation.