CasSR: Activating Image Power for Real-World Image Super-Resolution

TL;DR

CasSR introduces a cascaded diffusion approach with multi-attention to improve real-world image super-resolution, effectively reducing artifacts and enhancing detail recovery from degraded low-resolution images.

Contribution

The paper proposes a novel cascaded controllable diffusion model with multi-attention mechanisms for improved image super-resolution from severely degraded inputs.

Findings

Outperforms existing methods in qualitative and quantitative evaluations.

Produces more detailed and realistic high-resolution images.

Reduces artifacts and spurious content in super-resolved images.

Abstract

The objective of image super-resolution is to generate clean and high-resolution images from degraded versions. Recent advancements in diffusion modeling have led to the emergence of various image super-resolution techniques that leverage pretrained text-to-image (T2I) models. Nevertheless, due to the prevalent severe degradation in low-resolution images and the inherent characteristics of diffusion models, achieving high-fidelity image restoration remains challenging. Existing methods often exhibit issues including semantic loss, artifacts, and the introduction of spurious content not present in the original image. To tackle this challenge, we propose Cascaded diffusion for Super-Resolution, CasSR , a novel method designed to produce highly detailed and realistic images. In particular, we develop a cascaded controllable diffusion model that aims to optimize the extraction of information from low-resolution images. This model generates a preliminary reference image to facilitate initial information extraction and degradation mitigation. Furthermore, we propose a multi-attention mechanism to enhance the T2I model's capability in maximizing the restoration of the original image content. Through a comprehensive blend of qualitative and quantitative analyses, we substantiate the efficacy and superiority of our approach.