Improving Astrometric Precision with MLP-Driven Super-Resolution of Star Maps

Area of Science:

• Optics and Photonics
• Artificial Intelligence
• Aerospace Engineering

Background:

• Dynamic star simulators are crucial for star sensor calibration but suffer from positioning errors.
• Optical aberrations, assembly deviations, and device discreteness contribute to these errors.
• Existing methods struggle to comprehensively compensate for these error sources.

Purpose of the Study:

• To propose a super-resolution star map correction method for dynamic star simulators.
• To establish a data-driven framework for unified modeling and compensation of errors.
• To enhance the precision of star centroid positioning and inter-star angular distance.

Main Methods:

• A multi-layer perceptron (MLP) based super-resolution star map correction method is developed.
• A technical chain involving system calibration, aberration field modeling, and network correction is constructed.
• A data-driven, end-to-end framework is implemented for error compensation.

Main Results:

• The proposed method achieves sub-pixel accuracy in star centroid positioning.
• Maximum star centroid error is reduced by an average of 22.9%.
• Average inter-star angular distance error is reduced by 37.5%, outperforming traditional methods.

Conclusions:

• The MLP-based method offers a reliable approach for high-precision star map display.
• This technique provides significant improvements for star sensor ground calibration.
• The study demonstrates clear engineering application value in aerospace calibration.