Image inpainting is a method that reconstruct the missing area of an image by using the information in pixels of surrounding region. Its objective is to restore a degraded painting in such a way that the changes are not apparent to the observer. It is also an important step in many graphics algorithms. Image inpainting is an ill-posed inverse problem that has no well-defined unique solution.
To solve this problem, it is necessary to image priors. All methods are guided by the assumption that pixels in the known and unknown parts of the image share the same statistical properties or geometrical structures. This assumption translates into different local or global priors, with the goal of having an inpainted image as physically plausible and as visually pleasing as possible. The first category of methods, known as diffusion-based inpainting, introduces smoothness priors via parametric models or partial differential equations (PDEs) to propagate (or diffuse) local structures from the exterior to the interior of the hole. Many variants exist using different models (i.e. linear, nonlinear, isotropic, or anisotropic) to propagate in particular directions or to consider the curvature of the structure present in a local neighborhood. There also exists a method by using Gaussian process for image inpainting. But those methods are not suitable for recovering the texture of large areas. The second category of methods is based on the image statistical and self-similarity priors. It is more effective for the inpainting problem. The image gap is filled in recursively, inwards from the gap boundary. The statistics of image textures are assumed to be stationary or homogeneous. The synthesized texture is derived from similar regions in a texture sample or from the known part of the image. It is done by sampling, and by copying or stitching together patches taken from the known part of the image. These methods are called exemplar-based techniques.