Image Inpainting Using Curvature-Driven Diffusions Based on P-Laplace Operator

How to apply the curvature-driven diffusions model to image inpainting was discussed, then an image inpainting using curvature-driven diffusions based on p-Laplace operator was proposed. The algorithm mainly revert cracks and remove texts from photos. It makes full use of the nonlinear anisotropic diffusion of p-Laplace operator to inpaint damaged images. To test the models, we use standard peak signal to noise ratio (PSNR) and program runtime to quantify inpainting performance. Experimental results show that our algorithm´s PSNRs are larger than CDD´s and the runtimes are smaller; the proposed algorithm can improve inpainting speed and keep sharp edges; inpainting quality is improved.