Atmospheric correction of remote sensing imagery based on the surface spectrum’s vector space

Abstract

Due to the atmosphere effect, the qualities of images decrease conspicuously, practically in the visible bands, in the processing of earth observation by the satellite-borne sensors. Thus, removing the atmosphere effects has become a key step to improve the qualities of images and to retrieve the actual reflectivity of surface features. An atmospheric correction approach, called ACVSS (Atmospheric Correction based Vector Space of Spectrum), is proposed here based on the vector space of the features’ spectrum. The reflectance image of each band is retrieved first according to the radiative transfer equation, then the spectrum’s vector space is constructed using the infrared bands, and finally the residual errors of the reflectance images in the visible bands are corrected based on the pixel position in the spectrum’s vector space. The proposed methodology is verified through atmospheric correction on Landsat-7 ETM+ imagery. The experimental results show that our method is more accurate and the corrected image is more distinct, compared with those offered by current popular atmospheric correction software.