It’s necessary to remove thin cloud which is a common noise in remote sensing images. A wavelet removing cloud method that using frequency characteristics to seperately process wavelet coefficients of each layer has good cloud removing effect

but correct selection of some parameters has great effect on result

effective method has not been found yet. According to characteristics of remote sensing images

select evaluation index to construct fusion index

use this fusion indexes to find the best boundary layer and decomposition layers. By using this method to obtain two parameters

the goal for removing thin cloud is achieved

and the method is proved to be high efficient.