Present denoising algorithms of hybrid noise usually filter out Gaussian noise after removing pepper-noise and have many problems such as inaccurate image restoration and insufficient protection of image edge. In order to get rid of these problems, this paper presents an algorithm to remove the hybrid noise based on grid division. The algorithm introduces the definition of the edge grid cell and takes different methods to deal with the hybrid noise in edge grid cells and non- edge grid cells according to location of the noise. Experimental results show the proposed algorithm outperforms the existing algorithms in eliminating the hybrid noise and preserving the image details and has good performance for the single noise.