基于卫星云图的风矢场(云导风)度量模型与算法探讨

Research on the Measurement Model and Algorithm of Wind Vector Field Based on Satellite Cloud Images(Cloud Motion Winds)

对通过卫星云图来识别移动云块、计算风矢的问题进行了探讨.采用以点法式平面方程为核心的空间曲面方程组,实现了视场坐标变换;结合相邻帧差、中值滤波、阈值截取三种方法对卫星云图进行预处理,获得了云块分布;建立了滑动搜索图像匹配模型,并从数据挖掘以及图像特征的角度出发,设计了基于数据和基于统计参数这两种自适应算法,对匹配窗口大小、搜索范围进行优化;最后,为模型的继续改进提出了建议.

This paper discusses the issues about identifying moving clouds and calculating wind vectors by satellite cloud images. Firstly, a set of spatial surface equations with the equation of a plane in point-normal form as its core was adopted to realize the transformation between vision coordinates and latitude/longitude. Three methods, including consecutive frames subtraction, median filter algorithm and threshold value intercepting were combined to preprocess satellite cloud images, thus obtaining the distribution of clouds. Then, to match the cloud images, a sliding-window search model was established. And from the perspectives of data mining and image characteristic analyzing, two self-adaptive algorithms based on the data and statistical parameters were designed to optimize the sizes of the matching window and search scope. Finally, some suggestions to further improve the model were proposed.