卡尔曼滤波器在卫星遥感影像大地校正中的应用

在卫星遥感图像的预处理中,通过系统校正可以基本上消除影像内部的变形误差。然而,整幅影像的大地定位误差仍然是很大的,甚至几百米至1公里的数量级。本文介绍利用少量的地面控制点,采用递归算法而不是批处理算法,大大地提高大地定位精度。在单幅影像处理中,可以重新对卫星的轨道参数与姿态参数进行估值,从而提高影像的大地定位精度。在一条轨道上连续的多幅影像中,可以预报下一幅影像的大地位置,从而对那些不具有地面控制点的影像的大地校正中,实现外推计算。文中介绍并推导了卡尔曼滤波器方程中的转移矩阵和测量矩阵的系数,并推荐用数值回归的办法求得其数值解。文中最后介绍模拟试验及算法计算结果,并讨论其优缺点。

The Application of KALMAN FILTER in the Geodetic Correction of Satellite Remote Sensing Imagery

One of the image pre-processing tasks in the Remote Sensing Satellite Ground Station is the Systematic Correction Processing which may remove almost all the distortions inside the scene. However, the location accuracy is still rough and may probably reach hundred meters or kilometers. A recursive algorithm based on the Kalman Filter rather than the Batch Processing Algorithm using a few number of Ground Control Points will be used to enhance the location accuracy. Moreover, the ephemeric data and the attitude data will also be updated. In a single scene processing, 8 or 9 Control Points will be sufficient to reach the good result. In the continual multi-scene processing of a path, this algorithm may be propagated to predict the locations or the ephemeric and attitude data of the next scenes, in which the Control Point may be not available.In the section 2 of this paper, the recursive algorithm and the coefficients of the Transition Matrix and the Measurement Matrix in the Kalman Filter equations will be derived. The results of the simulation experiments including 3 Figures and 2 tables will be presented in the section 3. The section 4 has some discussions about the convergence and the advantages of this algorithm. Finally, 6 references will be given for the convenience in the future discussions.