3D Image Restoration using Diffusion and Fusion Techniques

We propose a novel method for restoring 3D blocks. The method is based on the Partial Differential Equations (PDE) framework and image fusion techniques. We first process a degraded block with two PDE based diffusion filters with different theoretical properties: a recent 3D directional diffusion based on the first order directional derivative and a new shock filter formulation. Adaptive wavelet-based fusion is then performed at an intermediate results level for combining the PDE effects. The procedure is iterated for a predefined number of steps and the restored block is obtained as the output of the last fusion step.