In this paper, we shall propose a lossless image compression method, in which we compress an image by saving the differences between the encoding pixels and their adjacent pixels. To deal with a pixel, we refer to not only its up pixel but also its left pixel and replace the value of the pixel itself with the value of difference. The experimental results show that the method gives quite an impressive performance. Since the values of most adjacent pixels in an image are close , we can use this simple, effective way to compress the image.