Explicit One-dimensional Radial Basis Function for Digital Image Inpainting

Abstract

Digital image inpainting is the process of repairing the damaged part of the image by filling that part using the information surrounding part. In this paper, we propose a new inpainting algorithm based on the heat equation. In our approach, the pixel value is analogous to temperature. So, the missing pixel values in the damaged parts are filled by diffusing the pixel values from surrounding areas. The spatial derivatives of the heat equation are discretized numerically using one-dimensional Radial Basis Function (RBF) method and the spatial derivative is discretized using explicit Euler method. Various numerical experiments on real images are conducted to show the accuracy of our algorithm.