A Thin Cloud Removal Method for Optical Remote Sensing Imagery Based on Spatial Variogram

Optical remote sensing images often suffer annoying contamination induced by clouds or aerosol, therefore cloud removal behaves an important work in the radiometric correction procedure for observed images. In this paper, the difference of spatial structural features between land surface and cloud noise is analyzed, and then an effective cloud removal algorithm is developed. This algorithm is based on the spatial variogram which can characterize the spatial correlation between image pixels and on the Kriging interpolation which can get unbiased optimal estimation to a regionalized variable. Preliminary test on the SPOT image indicates that the presented method performs well for removal of thin clouds. Some discussions about the applicability of the method and points of future efforts are given in the end.