Natural and Realistic Single Image Super-Resolution with Explicit Natural Manifold Discrimination

TL;DR

This paper introduces a novel super-resolution method that emphasizes naturalness and realism by constraining outputs within a natural manifold, leading to more perceptually convincing images.

Contribution

It proposes a new approach that enforces naturalness priors in the low-level domain to improve the realism of super-resolved images.

Findings

Better naturalness compared to recent algorithms

Enhanced perceptual quality with realistic textures

Maintains natural appearance without artifacts

Abstract

Recently, many convolutional neural networks for single image super-resolution (SISR) have been proposed, which focus on reconstructing the high-resolution images in terms of objective distortion measures. However, the networks trained with objective loss functions generally fail to reconstruct the realistic fine textures and details that are essential for better perceptual quality. Recovering the realistic details remains a challenging problem, and only a few works have been proposed which aim at increasing the perceptual quality by generating enhanced textures. However, the generated fake details often make undesirable artifacts and the overall image looks somewhat unnatural. Therefore, in this paper, we present a new approach to reconstructing realistic super-resolved images with high perceptual quality, while maintaining the naturalness of the result. In particular, we focus on the domain prior properties of SISR problem. Specifically, we define the naturalness prior in the low-level domain and constrain the output image in the natural manifold, which eventually generates more natural and realistic images. Our results show better naturalness compared to the recent super-resolution algorithms including perception-oriented ones.