Generally

there are two contents of detecting shadow areas and eliminating shadows in removing shadows of remote sensing images.In this study

the conclusion was drawn from the forming mechanism of remote sensing images and the reflection spectral characteristic of surface objects

which are the datum of green band having very high correlation with that of blue band

and the shadows of green band being obviously stronger than that of blue band.The reduced（divided） image derived from green band subtracted from（divided by） blue band

can be used to detect shadows with threshold value

due to the more differences between shadow areas and non-shadow areas in it.This method of detecting shadows is related to variational reflectance of the same object.In addition

the precision of detecting shadows with Histogram threshold value in near-infrared band is higher than the other’s.This method of detecting shadows is related to the reflectance of surface objects.Multispectral detecting shadow is the combination of the two methods.Shadow is sheltered by the objects higher than the ground from direct irradiance

the lost information is direct irradiance mainly

which is compensated

then the shadows could be removed.In this paper

a theoretical model

which was based on the radiation transfer theory to calculate the proportions of direct irradiance

diffuse irradiance and environment irradiance

was put forward to remove shadows through compensating direct irradiance.Finally

the method and the model were validated using the IKONOS images

and the results were turned up trumps.