Image inpainting using dynamic weighted kernels

Abstract

Image inpainting is a technique to automatically recover the damaged or missing regions in images. This paper presents a simple but fast approach for recovering the damaged regions of digital images. It uses a unique idea of dynamic weighted kernels. It proposes kernels of different sizes and weights to fill in the damaged regions with different width. The proposed technique has been tested for quality and execution time over a wide variety of images and results are found favorable.