TDDSR: Single-Step Diffusion with Two Discriminators for Super Resolution

2024-10-10

arXiv.org

AI Summary

This paper addresses the challenging single-step image super-resolution (SR) problem under unknown, complex degradations encountered in real-world scenarios. We propose an end-to-end, high-fidelity single-step SR method. Our key contributions are threefold: (1) a novel dual-discriminator architecture that enforces distribution alignment separately in both low- and high-resolution domains; (2) a learnable diffusion-based degradation model that explicitly captures the diversity of realistic degradations; and (3) a diffusion-prior-guided knowledge distillation strategy that enables a single-step denoising network to approach the reconstruction quality of multi-step diffusion models. Extensive experiments on real-world and face SR benchmarks demonstrate significant improvements over state-of-the-art methods: notable gains in PSNR and SSIM, over 100× faster inference speed, and reconstruction fidelity comparable to iterative diffusion-based approaches.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Multi-class/Multi-label Learning & Extreme Classification
• Search and Optimization: Learning to Search

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs

Abstract

Super-resolution methods are increasingly becoming popular for both real-world and face-specific tasks. Many existing approaches, however, rely on simplistic degradation models, which limits their ability to handle complex and unknown degradation patterns effectively. While diffusion-based super-resolution techniques have recently shown impressive results, they are still constrained by the need for numerous inference steps. To address this, we propose TDDSR, an efficient single-step diffusion-based super-resolution method. Our method, distilled from a pre-trained teacher model and based on a diffusion network, performs super-resolution in a single step. It integrates a learnable diffusion-based downsampler to capture diverse degradation patterns and employs two discriminators, one for high-resolution and one for low-resolution images, to enhance the overall performance. Experimental results demonstrate its effectiveness across real-world and face-specific SR tasks, achieving performance beyond other state-of-the-art models and comparable to previous diffusion methods with multiple sampling steps.

Problem

• Develops a co-learning framework for image super-resolution.
• Optimizes single-step diffusion upsampler and learnable downsampler.
• Enhances fidelity and perceptual quality in real-world SR tasks.

Innovation

• Co-learning upsampler and downsampler with diffusion
• Two discriminators enhance realistic degradation capture
• Single-step inference ensures efficient high-quality reconstruction