Continuous Degradation Modeling via Latent Flow Matching for Real-World Super-Resolution

2026-02-04

AI Summary

Existing super-resolution methods perform well under synthetic degradations such as bicubic downsampling but suffer significant performance degradation on real-world images due to complex, nonlinear distortions—including noise, blur, and compression artifacts—and the scarcity of real high–low resolution image pairs with limited scale diversity. This work proposes Latent Flow Matching, the first approach to incorporate flow matching into latent degradation modeling, enabling the continuous generation of realistic low-resolution images with arbitrary degradation levels from a single high-resolution image. By overcoming the conventional reliance on fixed degradation types and restricted scales, the method facilitates the construction of large-scale training data that substantially enhances the reconstruction quality of existing super-resolution models in real-world scenarios. Both quantitative and qualitative evaluations confirm its effectiveness.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Deep Generative Models & Autoencoders
• Natural Language Processing: Generation

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: Graph neural networks and deep learning approaches for Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

While deep learning-based super-resolution (SR) methods have shown impressive outcomes with synthetic degradation scenarios such as bicubic downsampling, they frequently struggle to perform well on real-world images that feature complex, nonlinear degradations like noise, blur, and compression artifacts. Recent efforts to address this issue have involved the painstaking compilation of real low-resolution (LR) and high-resolution (HR) image pairs, usually limited to several specific downscaling factors. To address these challenges, our work introduces a novel framework capable of synthesizing authentic LR images from a single HR image by leveraging the latent degradation space with flow matching. Our approach generates LR images with realistic artifacts at unseen degradation levels, which facilitates the creation of large-scale, real-world SR training datasets. Comprehensive quantitative and qualitative assessments verify that our synthetic LR images accurately replicate real-world degradations. Furthermore, both traditional and arbitrary-scale SR models trained using our datasets consistently yield much better HR outcomes.

Problem

• super-resolution
• real-world degradation
• image synthesis
• nonlinear degradation
• low-resolution image

Innovation

• Latent Flow Matching
• Continuous Degradation Modeling
• Real-World Super-Resolution
• Synthetic LR Generation
• Arbitrary-Scale SR