The Noise Reduction Algorithm for Star Detection

Abstract

Accurate star detection plays a critical role in star sensors for spacecraft attitude determination. However, various sources of noise such as cosmic radiation and sensor imperfections can degrade the accuracy of star centroid estimation. This paper presents a noise reduction algorithm that enhances the detection of star blobs in noisy star images. The proposed method integrates feature extraction through differential smoothing with adaptive thresholding techniques to effectively separate true star signals from background noise. To validate the algorithm, synthetic star images with different noise levels were tested. The results show that the proposed algorithm consistently detects all 20 target stars, achieving high detection accuracy even under severe noise conditions, outperforming conventional Laplacian of Gaussian (LoG) and Difference of Gaussian (DoG) approaches. This improvement in denoising leads to more precise centroid extraction, contributing to more reliable star sensor performance for space navigation applications.