Automatic correcting method of geometric distortion in meteorological satellite noaa image
Automatic correcting method of geometric distortion in meteorological satellite noaa image
复制标题
气象卫星noaa图像几何畸变自动校正方法
DOI：
10.1002/scj.4690200801
复制
发表时间：
1989
期刊：
影响因子：
--
通讯作者：
M. Onoe
中科院分区：
文献类型：
--
作者：
M. Takagi;T. Hiyama;M. Sone;M. Onoe
气象卫星 NOAA 图像中包含的几何失真已仅使用计算出的轨道值进行了校正。由于计算出的轨道值通常包含误差，因此已使用许多地面控制点（GCP）对其进行了精确修正。然而，如果NOAA-图像坐标中对应于GCP的位置被云覆盖，则由于该位置的位移不确定，因此需要通过选择另一个未被云覆盖的位置来手动进行校正。然而，如果需要手动校正的位置较多，除了位置位移会出现不同的值之外，还会增加手动操作的时间。本文提出了一种几何畸变的自动校正方法。该方法采用顺序相似性检测算法，使用包括控制点在内的海岸线数据进行模板匹配。对于此操作，有效地采用金字塔结构来提高处理速度。使用相关系数自动确定模板匹配的成功或失败。根据该结果计算仿射变换系数，从而自动校正几何畸变。地图坐标、地表坐标和NOAA图像坐标之间的变换采用四点插值法进行，缩短了处理时间。证明了该方法的有效性。
Geometric distortions contained in images of the meteorological satellite NOAA have been corrected using the calculated orbit values alone. Since the calculated orbit values usually contain errors, their precise corrections have been carried out using many ground control points (GCPs). If a position in the NOAA-image coordinate corresponding to a GCP is covered by clouds, however, the correction is carried out manually by choosing another position which is not covered by clouds, since the displacement of the position is not certain. If there are many positions to correct manually, however, the time of manual operations increases, in addition to the occurrence of different values in the displacement of the position. This paper proposes an automatic correcting method of the geometric distortion. The method employs the sequential similarity detection algorithm for template matching using coastal-line data including GCPs. For this operation, the pyramid structure is efficiently employed to increase the processing speed. A success or failure of template matching is determined automatically by using a correlation coefficient. From this result, the affine transformation coefficient is calculated so that the geometric distortion is corrected automatically. Transformations between the map coordinates, the ground-surface coordinates, and the NOAA-image coordinates are carried out by using the four-point interpolation method so that their processing time is shortened. The effectiveness of this method is demonstrated.