Image inpainting using frequency domain priors
This work aims to improve the quality of reconstructed high-frequency details in image inpainting, which is a problem for users requiring realistic and artifact-free image completion.
This paper addresses the challenge of reconstructing high-frequency details in image inpainting by incorporating frequency domain information (Discrete Fourier Transform) alongside spatial domain information. The proposed method, which includes a frequency-based deconvolution module, outperforms current state-of-the-art techniques qualitatively and quantitatively on datasets like CelebA, Paris Streetview, and DTD texture.
In this paper, we present a novel image inpainting technique using frequency domain information. Prior works on image inpainting predict the missing pixels by training neural networks using only the spatial domain information. However, these methods still struggle to reconstruct high-frequency details for real complex scenes, leading to a discrepancy in color, boundary artifacts, distorted patterns, and blurry textures. To alleviate these problems, we investigate if it is possible to obtain better performance by training the networks using frequency domain information (Discrete Fourier Transform) along with the spatial domain information. To this end, we propose a frequency-based deconvolution module that enables the network to learn the global context while selectively reconstructing the high-frequency components. We evaluate our proposed method on the publicly available datasets CelebA, Paris Streetview, and DTD texture dataset, and show that our method outperforms current state-of-the-art image inpainting techniques both qualitatively and quantitatively.