Image Enhancement and Restoration Approach Based on Anisotropic Diffusion

We propose a new approach for image enhancement, denoising and restoration, using an anisotropic diffusion based on P-M model and L.V and al. equation, replacing the gradient by motion by mean curvature to detect noise direction for each degraded pixel locally, applying the gradient in Gaussian kernel term to restore the degraded pixels and adding a time term supporting the restoration process.For execution progress, the numerical discretization for the terms of PDE modeling (obtained by the approximation by difference finite volumes finite method, Taylor method and Simpsons improved method), concludes an algorithm treats noised image regardless the noise type (salt-pepper or Gaussian or speckle) better than other filters based whether on anisotropic diffusion or total, shown in the experimental results (using MATLAB program), and demonstrated through PSNR and SSIM.