现有的云迹风提取方法只提供了对云覆盖的简单二值灰度估计.该文利用运动矢量估计的思想提出了一种云覆盖(云图)预测的新方法,这种方法可实现对卫星云图的灰度预测.这种新的方法是通过将一幅GMS云图分割成为许多大小相同的小块,并将它们分别匹配到一幅参考云图中以获取云的运动矢量,进而实现对卫星云图云覆盖的量化预测.进而还对该方法进行了优化,从而提高了算法运行速度.并对算法实现中出现的“马赛克”问题提出了解决方法,获得了令人满意的效果.

Traditional method of Cloud Motion Wind (CMW) provides only the binary gray-level prediction of cloud covering. In this paper, using the theory of motion vector estimation,a new method that can predict the true gray-level cloud covering is found. A GMS cloud image is divided into many small blocks with the same size in the method. These blocks are registered to a referenced image to get their motion vectors. By means of these vectors, true gray-level prediction of cloud covering is got. Furthermore, the method is optimized to improve the speed. The method to smooth“mosaic”is also provided, and satisfactory prediction results are achieved.