Daytime star spot detection based on spatial-temporal feature context enhancement for optical star sensors

Area of Science:

• Astronomy and Astrophysics
• Optical Engineering
• Computer Vision

Background:

• Optical star sensors are crucial for high-precision attitude determination in astronomical navigation.
• Traditional daytime star sensors face challenges due to intense atmospheric radiation, optical imperfections, and detector noise, reducing signal-to-noise ratio (SNR) and causing false alarms.
• Existing spatial feature-based methods are insufficient for effectively addressing these daytime operational constraints.

Purpose of the Study:

• To develop an advanced method for robust daytime star detection.
• To overcome the limitations of traditional methods in low SNR conditions and high false alarm rates.
• To enhance the operational reliability of star sensors in challenging optical environments.

Main Methods:

• A multi-stream feature fusion framework was proposed.
• Integration of a frame difference-optical flow guided spatial-temporal enhancement module.
• Utilization of attention mechanisms for noise suppression and motion-aware feature supplementation.

Main Results:

• The proposed approach demonstrated superior performance in daytime star detection.
• Validated through synthetic dataset training and real-world ground-based experiments.
• Significantly improved the signal-to-noise ratio for dim star spots and reduced false alarms.

Conclusions:

• The novel framework effectively enhances star detection in harsh daytime optical environments.
• The integration of spatial-temporal enhancement and attention mechanisms improves sensor robustness.
• This advancement enables more reliable all-time operation for astronomical navigation systems.