Provided are a method and apparatus for encoding/decoding a moving picture by using the diffusion properties of the image. In the method, a value of a current pixel is predicted from the values of neighboring pixels of the current pixel, based on the variations between the values of the neighboring pixels present in directions toward the current pixel; a correction value of the predicted value is calculated from at least one variation from among the variations; the value of the current pixel is determined by reflecting the correction value into the predicted value; and then, the original image is reconstructed using the determined value of the current pixel.