A Novel Image Interpolation Technique Based on Fractal Theory

Fractal interpolation can be used for scaling-up natural images, and it can retain the texture characters of the image well. But this approach must calculate the Hausdorff dimension first which is difficult to calculate by old ways, like Box-Counting approach and others. This paper proposes a new K dimension which is approximated to Hausdorff dimension and proposes an efficient way to calculate this dimension. Then we use the fractal Brownian motion (FBM) approach in the process of interpolation and get the post-processing images. In the end, the hardware architecture for fast implementation is proposed. Comparing to the other two traditional interpolation approaches, this approach can dramatically reduce the loss of the imagepsilas high-frequency components in scaling-up processing and easily applied in hardware design.