DeltaDiff: A Residual-Guided Diffusion Model for Enhanced Image Super-Resolution

2025-02-18

AI Summary

Existing diffusion models for image super-resolution inject global Gaussian noise in pixel space, leading to structural inconsistencies and detail distortion. To address this, we propose DeltaDiff—a residual-guided diffusion model that departs from conventional pixel-space noising and instead rigorously confines the entire diffusion process to the LR-HR residual space, modeling only essential high-frequency components. Our method employs a conditional diffusion architecture, leveraging a pre-trained backbone as a feature extractor; it replaces pixel-level denoising with residual prediction and introduces a progressive residual distillation strategy to enhance training stability. Extensive experiments demonstrate that DeltaDiff achieves state-of-the-art performance across multiple benchmarks, surpassing prior methods in PSNR, LPIPS, and perceptual quality. The code and pretrained models are publicly available.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Deep Neural Architectures and Foundation Models
• Search and Optimization: Learning to Search

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Responsible Web: Data and user privacy-enhancing technologies for the Web

Abstract

Recently, the application of diffusion models in super-resolution tasks has become a popular research direction. Existing work is focused on fully migrating diffusion models to SR tasks. The diffusion model is proposed in the field of image generation, so in order to make the generated results diverse, the diffusion model combines random Gaussian noise and distributed sampling to increase the randomness of the model. However, the essence of super-resolution tasks requires the model to generate high-resolution images with fidelity. Excessive addition of random factors can result in the model generating detailed information that does not belong to the HR image. To address this issue, we propose a new diffusion model called Deltadiff, which uses only residuals between images for diffusion, making the entire diffusion process more stable. The experimental results show that our method surpasses state-of-the-art models and generates results with better fidelity. Our code and model are publicly available at https://github.com/continueyang/DeltaDiff

Problem

• Enhances image super-resolution fidelity
• Reduces random noise in diffusion models
• Uses residual-guided diffusion for stable results

Innovation

• Residual-guided diffusion model
• Enhanced image super-resolution
• Stable diffusion process