A local nontexture image inpainting and denoising based on nonlinear PDEs

Image inpainting is an interpolation which guesses and fills in an information-losing portion according to its surroundings. Image inpainting belongs to the category of image restoration and reconstruction. The Total Variation model for image inpainting is an effective approach, however, it fails to process the images damaged by a big noise and choosing its regular parameters is also difficult. We have made some improvements on it, which can simultaneously denoise and preserve edges effectively filling in the missing area. This has been indicated theoretically and experimentally.