Variational image inpainting models based on p-harmonic energy functional

We address the problem of image inpainting with variational methods. From the study of existing formalisms, two variational inpainting models based on p-harmonic energy function are proposed. The associated Euler-Lagrange equations lead to p-harmonic equations. We analyze the physical characteristics of the p-Harmonic equations in local coordinates and explain that diffusion behavioral of p-harmonic is superior to that of total variation model in essence. Then proper numerical schemes are designed to handle their computation. They are finally applied on a wide variety of image processing problems, including color image restoration, inpainting and zooming. All these results show that the proposed models are effective.