R. Terebes, M. Borda, C. Ludusan, and S. Pop (Romania)
Diffusion, fusion, shock filters, orientation.
This paper presents a new image restoration method based on the Partial Derivatives Equations (PDE) framework and image fusion techniques. The degraded image is processed independently using two PDE filters -a tensor based directional diffusion filter and a scalar shock filter than the results are combined through fusion, the final processed image being obtained by a recurrence of diffusion and fusion steps. Fusion takes place at an intermediate results level through a weighted averaging/selection fusion rule. Preliminary results show such an approach can be used with success for processing noisy, blurred images. An experimental setup involving synthetic and real images is used in the last section to illustrate the efficiency of the method.
Important Links:
Go Back
