The purpose of this paper is to present a novel algorithm for image inpainting, which has been widely used for removing unwanted objects from images or reconstructing damaged photographs.
An image piecewise inpainting technique based on radial basis function (RBF) is used to transform the 2D image inpainting problem into 3D implicit surface reconstruction problem. And a RBF center reduction method is proposed. By RBF resampling, the algorithm can nicely fix the damaged image or remove specific objects.
Experimental results show that the proposed algorithms can prevent the edge blur caused by the isotropic character of RBF, and effectively reduce the RBF centers without a loss in accuracy.
The proposed inpainting approach is interesting for its combination of RBF method and region segmentation that can handle the restoring of high‐variation areas.
