Pde based inpainting algorithms: performance evaluation of the Cahn-Hillard model

Image inpainting consists in restoring a missing or a damaged part of an image on the basis of the signal information in the pixels sur- rounding the missing domain. To this aim a suitable image model is needed to represent the signal features to be reproduced within the inpainting domain, also depending on the size of the missing area. With no claim of completeness, in this paper the main streamline of the development of the PDE based models is retraced. Then, the Cahn-Hillard model for binary images is analyzed in detail and its performances are evaluated on some numerical experiments.