Learning Structral coherence Via Generative Adversarial Network for Single Image Super-Resolution

TL;DR

This paper introduces a GAN-based approach for single image super-resolution that incorporates a gradient branch and a U-net discriminator to enhance structural coherence and detail realism, outperforming existing methods.

Contribution

It proposes a novel GAN architecture with a gradient branch and U-net discriminator to improve structural and perceptual quality in super-resolution tasks.

Findings

Outperforms state-of-the-art perceptual-driven SR methods in perception index (PI).

Produces more geometrically consistent textures.

Generates more natural and visually pleasing details.

Abstract

Among the major remaining challenges for single image super resolution (SISR) is the capacity to recover coherent images with global shapes and local details conforming to human vision system. Recent generative adversarial network (GAN) based SISR methods have yielded overall realistic SR images, however, there are always unpleasant textures accompanied with structural distortions in local regions. To target these issues, we introduce the gradient branch into the generator to preserve structural information by restoring high-resolution gradient maps in SR process. In addition, we utilize a U-net based discriminator to consider both the whole image and the detailed per-pixel authenticity, which could encourage the generator to maintain overall coherence of the reconstructed images. Moreover, we have studied objective functions and LPIPS perceptual loss is added to generate more realistic and natural details. Experimental results show that our proposed method outperforms state-of-the-art perceptual-driven SR methods in perception index (PI), and obtains more geometrically consistent and visually pleasing textures in natural image restoration.