关键词: 厚云, 云阴影, 归一化差值, 线性回归分析

Abstract:

Previous method of cloud and its shadow removal in the remote sensing image focus on the geometrical characteristic recovery of coverd terrain,this method is against the information extraction  based on the spectral characteristics.A method based on the spectral characteristics of covered terrain category reconstruction was  proposed.Adopt two adjacent years phenophase remote sensing data,use normalized difference value and decision tree classification method to detect the thick cloud and its shadow,and extract corresponding regional surface coverage type information,in remote sensing image of two various bands,do the linear regression analysis to each surface feature of the covered terrain,to obtain the reconstructed data of the covered terrain.The experimental results show that the method can recover the geometry information of the covered terrain，and it also realizes its spectral information simulation.

Key words: thick clouds, shadow, normalization, linear regression analysis