Improved algorithm for removing thin cloud in single remote sensing image

Abstract

Because the algorithm of cloud threshold often generates boundary effect,this paper proposed an improved algorithm based on wavelet transform and homomorphic filter.The image with cloud was decomposed by wavelet transform to find the proper number of demarcation levels.Cloud could be removed by making homomorphic filtering to the higher level coefficients,while giving the lower level detailed coefficients and the approximation coefficients some weight factors respectively.The three parts of coefficients were reconstructed and fused to get processed result.The experimental results indicate that the proposed algorithm can remove the thin cloud cover effectively,maintain the details better and prevent producing the boundaries.