A novel initial image for fast fractal decoding

In order to speed up the fractal decoding, a novel initial image is proposed in this paper. In the case of the same decoding method, the convergence speed depends heavily on the similarity of the initial image with respect to the original image. With the development of the fractal coding, the matching errors between the range blocks and domain blocks become smaller and smaller. Thus, with the fractal code obtained in the fractal encoding process, we can reconstruct a novel initial image that can approximate the original image better. Experiments show that the initial image we propose can provide faster convergence speed than the conventional initial images.