The conversion of sensor the Top of the Atmosphere Signals (TOAS), to reflectivity representing land surface target characters is an important sector in remote sensing information quantification. Assuming atmosphere disturbance being excluded, because of Adjacency Effect among pixels, the DN value of a target pixel contain the scatter contributions of background pixels. The magnitude of the contribution is directly proportional to the background pixel reflectivity and is correlative with the distance to target pixel. Adjacency Effect leads to blurry surface edges and low contrast in remote sensing images, which result in the uncertainty of remote sensing information and difficulty of quantitative explanation of surface features. Along with the increase of satellite spatial resolution and remote sensing information quantification degree, Adjacency Effect and its correction methods are researched to restore images. Based on radiation transfer model, the Adjacency Effect in a scene of Landsat5 TM data of Beijing (as an example) is analyzed and corrected. A significant improvement of the image quality was achieved.