Structure-Preserving Super Resolution with Gradient Guidance

TL;DR

This paper introduces a structure-preserving super resolution method that uses gradient guidance to enhance geometric structures in images, improving perceptual quality while maintaining competitive PSNR and SSIM.

Contribution

It proposes a gradient-based approach with a gradient branch and gradient loss to better preserve structures in GAN-based super resolution.

Findings

Achieves top PI and LPIPS scores among perceptual SR methods.

Maintains comparable PSNR and SSIM to state-of-the-art methods.

Demonstrates superior structural restoration in visual results.

Abstract

Structures matter in single image super resolution (SISR). Recent studies benefiting from generative adversarial network (GAN) have promoted the development of SISR by recovering photo-realistic images. However, there are always undesired structural distortions in the recovered images. In this paper, we propose a structure-preserving super resolution method to alleviate the above issue while maintaining the merits of GAN-based methods to generate perceptual-pleasant details. Specifically, we exploit gradient maps of images to guide the recovery in two aspects. On the one hand, we restore high-resolution gradient maps by a gradient branch to provide additional structure priors for the SR process. On the other hand, we propose a gradient loss which imposes a second-order restriction on the super-resolved images. Along with the previous image-space loss functions, the gradient-space objectives help generative networks concentrate more on geometric structures. Moreover, our method is model-agnostic, which can be potentially used for off-the-shelf SR networks. Experimental results show that we achieve the best PI and LPIPS performance and meanwhile comparable PSNR and SSIM compared with state-of-the-art perceptual-driven SR methods. Visual results demonstrate our superiority in restoring structures while generating natural SR images.