Exemplar-based image inpainting based on pixel inhomogeneity factor
Qinhui Fan, Hongsheng Liu, Zhizhong Fu, Xiaofeng Li · 2017
In the exemplar-based image inpainting process, the filling order, which is decided by a predefined priority function, is crucial to the image inpainting result. The traditional exemplar-based image inpainting methods usually generate wrong geometry in presence of strong structures in the damaged image. In this paper, we present a novel priority function based on pixel inhomogeneity factor. Compared with the classical and state-of-the-art exemplar-based inpainting approaches, the proposed method enables better inpainting of geometry and texture. Experiments on synthetic and natural images show the advantages of the proposed approach.