The Measurement Technology Study Of Star Tracker With Wide Field Of View And Small Aperture

Modern satellites which present the trend of lightweight, miniaturization, low cost and high functional density, have become an important part of space system gradually in recent years. As the core component to achieve autonomous attitude measurement of a small satellite, star tracker has board application prospect with the advantage of good autonomy, high accuracy and functional reliability.To fit in the needs of modern satellite technology, star tracker presents a tendency towards wide field of view, small aperture and high efficiency. However, there are still lots of technical difficulties of star tracker measurement algorithm, in-depth study has a very important significance.In this paper, according to the function demands of the actual system, the main research work and innovation points are as follows.Star image pretreatment mainly refers to star image denoising and star points extraction and location. The morphological opening operation is chosen to denoise star image based on the analysis of noise feature. After comparing the efficiency between traditional connected domain algorithm and cross projection algorithm, we extract the star points with an improved cross projection algorithm. In order to improve location precision, a sub-pixel locating technology is applied.The star sensor’s calibration is important for the high precision of star tracker. Against the shortcomings of traditional calibration methods such as distortion models method and directly mapping method, we used trained BP neural network to realize lens distortion correction and introduced genetic algorithm to improve the BP neural network algorithm. Experimental results shows that our method efficiently improve the accuracy with good stability.Compare and summarize regular star pattern recognition algorithms. An improved algorithm based on star inter-angle triangle algorithm is proposed. We used two star inter-angle triangles with common edge as the basic matching element to reduce the redundancy of matching stars and decrease the influence of fake stars. In addition, we introduce k-d tree to build up the pattern database, which speed up the star pattern matching.With the observation window applied by the star tracker with wide field of view and small aperture, we analyzed the ability to detect space moving target. Using the motional differences between stars and moving target, we can get real moving target. Experimental results that multiple targets can be detected and the algorithm is of robustness.