中文摘要：

由于光学成像系统的非线性几何畸变，使得星敏感器所获星图与理想星图有一定的差别，为完成高精度定姿，必须对畸变星图进行校正。首先，介绍了光学系统的畸变原理，并建立了以径向几何畸变为主的星图畸变模型。之后引入改进遗传算法对畸变参数进行了优化：采用引进成长算子的二进制编码，避免了算法陷入伪极值；通过改进适应度函数和适时调整变异概率，避免了计算中产生的早熟收敛问题。通过与基本遗传算法的比较结果表明，该方法不仅能够降低75．3％的相对误差，而且还提高了16ms的畸变校正速度，基本能够满足星图识别和姿态确定对精度高、实时性强等性能的要求。

英文摘要：

Due to the non-linear geometric distortion in optical imaging system, the difference exists between the real star images and ideal images. According to precisely fulfill the attitude detection, the distortion must be calibrated. Firstly, based on the pretreatment to the star image, a gray weighted centroid method is applied to get the star point coordinates. The distortion principle of the optical system is presented and the distortion model that considers the radial geometry distortion as primary factor is established. Then an improved genetic algorithm （GA） is proposed to optimize the distortion parameters. A growing operator applied to binary-coded is introduced to prevent the algorithm from falling into the pseudo-extremum ; the fitness evaluation function is improved and the mutation probability is adjusted timely to avoid premature convergence of genetic algorithms. By comparing with the traditional GA, the results indicate that this method not only reduces 75.3% of the relatively errors, but also upgrade the speed of distortion calibration with 16ms. It can meet the capability requirements of star map identification and attitude detection with higher precision and rapid speed as well as strong real time performance.