In terms of background reflectance computation and true reflectance acquisition, the two adjacency effect correction algorithms are studied comparatively. One is based on three-dimensional radiative transfer model SHDOM (Spectral Harmonics Discrete Ordinary Method) empirical equation for the nadir-viewing optical remote sensing image and the other is based on the synchronically measured spectral data. The data sources are Landsat-5 TM image with synchronized measured spectral data of sand, and ASTER image with synchronized measured spectral data of grass. The results demonstrate that the adjacency effect correction algorithm based on SHDOM empirical equation has more profound theoretical foundation according to the radiative transfer theory, and the image corrected by the adjacency effect correction algorithm based on the synchronized measured spectral data looks better.