中文摘要：

获取高精度事后姿态数据是提高遥感平台成像质量的必要条件之一,离线处理可有效降低敏感器测量误差,从而获得更高的姿态确定精度。基于滤波的校正方法中,星敏感器低频误差（LFE）与陀螺漂移将产生耦合影响导致校正精度低,本文针对该问题推导了耦合误差的数学模型,并设计了一种两步双向平滑事后处理算法,将陀螺漂移与低频误差分两步校正,通过反复滤波剥离陀螺漂移与低频误差。同时,针对低频误差参数收敛速度慢、噪声参数调节困难的问题,利用一种基于极大似然估计（MLE）的固定窗口自适应双向滤波算法进行处理以获得更好的噪声估计,提高了收敛速度和收敛精度。文中仿真工况下,离线姿态确定精度可达到0.8″（3σ）,低频误差参数完全收敛时间不超过4个轨道周期。

英文摘要：

High-accuracy post attitude data is critical to the improvement of image quality of remote sensing platforms. Dur- ing offline processing, errors of attitude sensors can be efficiently calibrated to achieve higher precision of attitude determi- nation. However, coupling influence of low frequency error （LFE） and gyroscope drift can cause the decrease of calibration precision. In order to solve the problem, a mathematical model of the influence is derived in this paper. Meanwhile, a two- step bidirectional smoothing algorithm is proposed to calibrated separately gyroscope drift and LFE. Gyroscope drift and LFE can be perfectly separated with the proposed method. In order to solve the problems of slow convergence of LFE parameters and the difficulty of tuning noise parameters a maximum-likelihood-estimation （MLE） based bidirectional adaptive filtering al- gorithm is developed, which can improve both convergence speed and precision dramatically. Under the simulation condition in this paper, the accuracy of offline attitude determination reaches 0.8＂（3 σ） and the convergence time of LFE parameters is not more than 4 orbital periods.