Skip to Main Content
In fractal coding technique, an image is encoded by making use of its self-similarity property. The image can be reconstructed with some well-defined contractive mappings based on this property and hence, in theory, the reconstructed image can be of any desirable size by using an initial image of appropriate size during the decoding process. However, in practice, the enlarged image is always degraded due to the sub-optimal contractive mappings used. In this paper, a fractal-based image enlargement technique is proposed to reduce this problem. This technique can preserve the details in edge regions while maintaining the smoothness in flat regions, which is superior to conventional image enlargement techniques such as bilinear interpolation and cubic convolution.