In this paper, I propose a method for efficient coding of images using cellular automata. This method allows us to describe each selected group of neighboring cells of bend points in the contour. This method enables us to compress the image code. These groups will be separated to objects in an image by using cellular automata which uses bend-points determination.