Research On Star Map Image Processing Algorithm Aiming At Attitude-estimation Improvement Of Star Sensor

As an important attitude-estimation equipment,star sensor is widely used in satellites due to its high accuracy,autonomous navigation ability and robustness to interference.With the rapid growth of aerospace activities,the requirement of attitude-estimation accuracy of spaceflight has become stricter,which challenges the technology of star sensor.For star sensors,the attitude information of spacecraft is acquired by computing the star centroid of a star map.However,the star centroid accuracy of a star sensor is sensitive to certain factors inside and outside star sensor components,resulting a grave attitude-estimation error.For the development of star sensor with higher attitude-estimation accuracy,it is important to eliminate the noise in star maps and improve the star centroiding accuracy of star sensor.This research aims at star image processing algorithms for star tracker attitude-estimation accuracy improvement.The research of this paper consists of five parts mainly.Part 1 is the introduction of star sensor and the main factors which constrain the accuracy of attitude-estimation of star sensor.An analysis of the star centroid accuracy is followed.Star map background noise and Single Event Effect noise elimination methods at home and abroad are sentenced in the end.Part 2 firstly analyzes the characteristic of star images captured by star sensor.The category and formation of star map background noise are analyzed behind.According to the analysis of the physics of Single Event Effect,a gray scale response model of noisy star image under Single Event Effect is proposed based on its formation and characteristic in star map.The influence of Single Event Effect on star centroid is analyzed behind.Part 3 is the research on star map background noise elimination.According to the principle of mathematical morphology,mathematical morphology filter can be used in star map denoising.In order to determine the shape and size of structure element used in mathematical morphology filtering on star map,a contrast experiment is carried out.The proposed mathematical morphology filtering method is certified by simulation experiment.The capability of proposed method under different noise level has also been tested.Part 4 is the research on Single Event Effect noise elimination,aiming at the bright noise pixels within dispersed region of a star image.Based on the gray scale response model of noisy star image under Single Event Effect and the characteristic of star image,a reference model using adjusted point spread function is constructed.The parameter estimation method of proposed PSF reference model is studied.The proposed PSF reference model together with gaussian side window filter are utilized for star image correction.Simulation experiment is carried out to test the proposed method.Part 5 is a combination experiment towards star map background noise and Single Event Effect noise.Both actual star map and simulated star map are used to test the proposed method.Experiments towards the capability of proposed method under different noise level are carried out.In order to simulate the actual working condition of star sensor,an actual star map and a simulated star map are combined.The result shows that the signal to noise ratio of this combined star map is enhanced after using the proposed method.