Image　magnification　plays　an　important　role　in　many　fields.　An　algorithm　based　on　fractal　codes　and　model　constraint　is　proposed　in　this　paper.　The　fractal　codes　can　carry　the　spatial　information　as　well　as　the　self-similarities　presented　in　the　image.　First,　the　image　is　magnified　with　multiple　range　block　partition　schemes,　then　their　results　are　averaged　to　get　one　magnification　image　and　thus　the　block　artifact　is　reduced.　Finally,　the　magnification　image　is　modified　by　a　constrained　model.　Experiments　show　that　the　proposed　approach　performs　better　than　the　other　fractal　based　magnification　method.