Motion-Aware Restoration of Smeared Star Tracker Images for High-Performance Satellite Attitude Determination

Yi-Chen Tseng and Shau-Shiun Jan

Peer Reviewed

Abstract:	Precise attitude determination is essential for satellite missions; however, traditional sensors are often limited by drift, reliance on external references, and low accuracy. Although star sensors have high precision, dynamic conditions such as spacecraft rotation or vibration can cause blurring of their star images, reducing centroiding accuracy and compromising mission reliability. This paper presents an integrated framework that combines angular velocity estimation with image restoration to enable the effective correction of motion-blurred star images. Dynamic star fields were simulated in this study by using the SKY2000 catalog, with the simulation incorporating parameters such as field of view, focal length, pixel size, stellar magnitude, exposure time, and angular velocity. Star motion blur was modeled using line integrals along trajectories to produce physically accurate data for algorithm validation and verification. Key motion parameters—including blur direction, star trail length, and intensity distribution—were extracted using correlation filtering, mathematical morphology, and image differentiation. These parameters were applied within two complementary estimation methods: an optical flow approach for fine motion detection and a geometric method utilizing star trail chords and normal features. The fusion of these methods through a Kalman filter ensured robust estimation of angular velocity across a wide dynamic range. Crucially, the estimated angular velocity enabled derivation of the point spread function (PSF), which accurately represents motion-induced blur. Incorporating this PSF into the Lucy–Richardson deconvolution algorithm yielded sharp, near-Gaussian star profiles, considerably enhancing the signal-to-noise ratio and centroiding precision. Attitude determination results obtained before versus after restoration were compared, and the comparison revealed that the proposed framework achieved substantial improvements in estimation accuracy, providing a unified, physicsbased solution for reliable and high-performance star tracking during dynamic satellite missions.
Published in:	Proceedings of the 2026 International Technical Meeting of The Institute of Navigation January 26 - 29, 2026 Hyatt Regency Orange County Anaheim, California
Pages:	717 - 725
Cite this article:	Tseng, Yi-Chen, Jan, Shau-Shiun, "Motion-Aware Restoration of Smeared Star Tracker Images for High-Performance Satellite Attitude Determination," Proceedings of the 2026 International Technical Meeting of The Institute of Navigation, Anaheim, California, January 2026, pp. 717-725. https://doi.org/10.33012/2026.20519
Full Paper:	ION Members/Non-Members: 1 Download Credit Sign In