数学形态学对卫星云图特征量的提取分析

Extraction and Analysis of Feature Quantities of Satellite Cloud Images Based on Mathematical Morphology

摘要: 本文根据数学形态学的原理，用数学形态学的方法和MATLABLE编程实现了卫星云图形态及特征量的提取。并用主成分分析对四类云系进行了很好的识别：涡旋云系短轴较均匀，像素面积大，旋转力度较大，反映到欧拉数的数目很多；条状云系长轴较长，像素集中反映到连通域数目较少；絮状云系短轴均匀，方向各异；层状云系长短轴相差不大，反映在体态比值较小确均匀。

Abstract: Based on the principle of mathematical morphology, this paper uses the method of mathematical morphology and MATLAB programming to extract the morphology and feature quantities of satellite cloud images. Principal component analysis is used to identify four types of cloud systems very well: the short axis of vortex cloud systems is relatively uniform, the pixel area is large, and the rotation force is relatively large, which is reflected in a large number of Euler numbers; the long axis of strip cloud systems is relatively long, and the pixels are concentrated, which is reflected in a small number of connected domains; the short axis of cotton-like cloud systems is uniform, and the directions are different; the long and short axes of stratiform cloud systems are not much different, which is reflected in a small and uniform body ratio.