一种高保真同态滤波遥感影像薄云去除方法

A High-fidelity Method of Removing Thin Cloud from Remote Sensing Digital Images Based on Homomorphic Filtering

同态滤波是一种常用的遥感影像薄云去除方法,但传统方法在抑制薄云对应的低频区域时,会不可避免地改变非云区的辐射信息。本文提出一种高保真遥感影像薄云去除方法,在同态滤波框架内利用基于区域模板的检测方法进行云区判别,处理时仅对云区用同态滤波结果进行替换,无云区则保留原始影像的亮度值。为进一步提高影像的保真度,影像拉伸处理时进一步剔除了云区对拉伸系数的影响。实验结果表明,本文方法能够在有效去除薄云的同时保留非云区的辐射信息,具有高保真特性。

The traditional homomorphic filtering frequently used in the cloud-removing has effect on the cloud in low-frequency region,but the traditional method used to remove thin cloud in the area of low frequency inevitably changes the region information.This article suggests a high-fidelity method of removing thin cloud from digital remote sensing images,which is based on the methods of regional template for the district judge after homomorphic filtering.After filtering only replace the cloud-region to the result of homomorphic filtering but un-cloud region still keeps the original data.To further enhance the image fidelity,it eliminates the cloud-region influence in the further process.The experimental results show that this method can effectively remove thin cloud and keeps the high fidelity.