Gradient-Based Search Space Reduction for Fast Exemplar-Based Image Inpainting

In this paper a fast exemplar-based imageinpainting method is presented by a gradient-based search spacereduction. Our method works based on the priority functiondefined by Criminisi et al. [5]. The priority function encouragespixels with high gradient magnitudes to be selected on imagecompletion path. The behavior of patches which have highgradients at their center are heavily impressed by the gradientvector. In this paper we have employed the gradient vector toreduce the search space and speed up the image inpaintingalgorithm. By forming lists of pixels with similar gradient vectorsat preprocessing stage and selecting the most relevant list foreach patch, the search space is reduced in the order of 100 times.As illustrated in the experiments the proposed method speeds-upthe original inpainting algorithm significantly.

Keywords:

Image inpainting , texture synthesis , imagecompletion , exemplar , search space reduction