Star Identification Algorithm Based on Multi-Dimensional Features and Multi-Layered Joint Screening for Star Sensors

The algorithm for star identification is a crucial technology for determining the orientation of spacecraft using star sensors. Traditional star identification algorithms achieve matching by seeking a unique or a few optimal solutions. However, in high-noise environments, some solutions may be lost, which could result in matching failure. A new lost-in-space architecture algorithm aimed at rapid identification under high star position noise conditions by directly using star positions for final matching is proposed in this paper. The main idea of this algorithm is to construct sufficiently redundant navigation triangles, fully utilizing the physical relationships of the features and forming a screening method from high to low dimensions and from loose to strict. During identification, a multi-layer joint screening matching method is adopted to screen triangles as a whole, narrowing the range of matches quickly while retaining error tolerance. In a series of simulation experiments, this algorithm achieved identification rates of 99.51%, 99.06%, and 98.42% for 2.0 pixel star position noise, 1.0 Mv star magnitude noise, and 5 false stars, respectively. In terms of practical application, all 1000 star images taken by the star sensor in orbit have been successfully identified, and it only takes 28ms to identify each image. In addition, star images taken by consumer-grade cameras from the ground also show that the algorithm has strong robustness to star position noise, magnitude error and false star interference in more severe environments. This method provides partial algorithmic reference for non-specialized design of star sensors for low-cost, large-scale satellites in the future.