Research On Star Image Simulation And Star Extraction Method For CCD's Star Tracker

With the development of application technology of moonlets, CCD's star tracker with high measure precision is popular. Rapid and reliable star pattern recognition algorithm is the most important part of control system. Because the cost of spaceflight experiment is very high, it is impossible to acquire real-time star images in the sky in the course of debugging star tracker on the ground and simulating software algorithm. Thus in order to debug and estimate star pattern recognition algorithm, we need to simulate real-time star images from star tracker in the ground with a computer. At present, plentiful research focuses on star pattern recognition algorithm. However, the research on star image simulation and star extraction from star image is much fewer. So the main research of this article focuses on star image simulation and star extraction. Firstly, the article introduces the development situation of star image simulation and star extraction from star image, and then discusses some aspects which may have effects on star image simulations, including coordinate transform from celestial sphere to image plane, the transform from brightness to gray, defocus modeling, nebula and star cluster, cosmic background radiation, CCD's self-noise, image's geometrical aberration, image blurring due to motion. According to the relation between brightness and gray, the transform expression from brightness to gray is built. Simulation parameters have been set in detail,and the simulation results have been provided. Nebula and star cluster are simulated and their simulation results are provided accordingly. Considering practical parameters of secondary planet and interval time of each star image, whether motion blur would happen is discussed, and the simulation results of motion blur are also provided. Besides single star image simulation, we also simulate movement process of continuous star images with orbit expressions. In addition, based on the similarities between star and small target,a new method of star extraction from star image comprised of high pass filter and dynamic threshold is proposed to deal with the star acquisition. The star image is filtered by the high pass filter with dynamic threshold to obtain possible star pixels. Owing to the abundance of illusive objects, the possible star pixels are acquired by using a minimum threshold fixed before-hand. The simulation results illustrate that the vector technique and our algorithm have same precision with the absence of noise. Once noise is added onto star image, the precision of proposed method is much higher than that of the vector technique, while the processing efficiency of them remains the same. Thus the proposed method has strong anti-noise ability.