So we coding the image with its relativity: Coding the first subimage singly and then coding the differences between the first subimage and the rest three.