Algorithm for removing thin cloud from remote sensing digital images based on homomorphic filtering

The traditional homomorphic filtering frequently used in cloud-removing has effect on the clouds in low-frequency region, but is not effective for those in high-frequency region, because of the limitation of high-pass filters it used. Based on the study of the traditional homomorphic algorithm, the most effective high-pass filter in frequency domain is adopted. And then, in order to remove the cloud in high-frequency region, a popular nonlinear spatial filter, 3×3 median filter, is introduced, which replaces the value of the central pixel by the median of the gray levels in the 3×3 neighborhood of that pixel. At last, analysis of the experiment results in vision and statistics is carried out, which shows that the improved method is effective in removing the cloud in high-frequency region