利用海岸线的海洋遥感图像控制点(GCP)自动匹配法

Automatic location of GCPs at sea

研究一种利用海岸线自动确定海上地面控制点（ＧＣＰ）的方法．借助边缘检测技术提取海岸线，将海岸线的点全部作为控制点的备选集合，利用相关松弛法寻找同名点，建立一种可靠判别机制来保证海岸线上ＧＣＰ的正确性，通过平均法和插值法求取海上ＧＣＰ的值．利用本方法可以方便地得到分布密集的ＧＣＰ值，其中海上ＧＣＰ值具有子像元精度，同时研究了一种新的几何变换方法，即用插值法直接求取需校正图像的几何变换坐标，随着插值密度的增加，求得海上ＧＣＰ值的点就越多，最后使所有点的坐标值通过插值方法计算得到，代替二元ｎ次方程组进行空间坐标变换，使遥感图像的几何配准误差在某种程度上达到０．利用该方法对海洋遥感资料进行几何配准，可以提高配准的精度和节省机时，为遥感资料的动态监测和数据库建设创造了有利条件。

A new method is developed to locate ground control points (GCPs) at sea. All points at the coastline, extracted by Roberts edge operator from the first image, are taken as candidate points of GCPs. The counterpart points in the second image are found by correlation-relaxation (CR) technique. The accuracy of GCPs is guaranteed by a decision rule. The methods of average and interpolation are used to locate new GCPs at sea. A new transformation of geometric rectification is also developed with zero RES error. The coordinates of the rectified image can be directly obtained from interpolation method instead of a 1st-order bivariate transformation. The above-mentioned approach can be used in dynamic monitoring and integrating of satellite data.