mlsanityData to 14 Oct 2026

From Zero to Detail: A Progressive Spectral Decoupling Paradigm for UHD Image Restoration with New Benchmark

arXiv 2604.15654 ↗First posted 17 Apr 2026cs.CVcited by 7

Not on the map. Why?

Abstract (arXiv)

Ultra-high-definition (UHD) image restoration poses unique challenges due to the high spatial resolution, diverse content, and fine-grained structures present in UHD images. To address these issues, we introduce a progressive spectral decomposition for the restoration process, decomposing it into three stages: zero-frequency enhancement, low-frequency restoration, and high-frequency refinement. Based on this formulation, we propose a novel framework, ERR, which integrates three cooperative sub-networks: the zero-frequency enhancer (ZFE), the low-frequency restorer (LFR), and the high-frequency refiner (HFR). The ZFE incorporates global priors to learn holistic mappings, the LFR reconstructs the main content by focusing on coarse-scale information, and the HFR adopts our proposed frequency-windowed Kolmogorov-Arnold Network (FW-KAN) to recover fine textures and intricate details for high-fidelity restoration. To further advance research in UHD image restoration, we also construct a large-scale, high-quality benchmark dataset, LSUHDIR, comprising 82,126 UHD images with diverse scenes and rich content. Our proposed methods demonstrate superior performance across a range of UHD image restoration tasks, and extensive ablation studies confirm the contribution and necessity of each module. Project page: https://github.com/NJU-PCALab/ERR.

Key facts. Cited by 7 papers in the map (data through 14 Oct 2026). Builds on 89 of its 130 references that are in the map. First posted on arXiv 17 Apr 2026.

Builds on

89 of 130 references.

References · 89

  1. 2018The Unreasonable Effectiveness of Deep Features as a Perceptual Metriccited by 11,177
  2. 2016SGDR: Stochastic Gradient Descent with Warm Restartscited by 5,045
  3. 2021SwinIR: Image Restoration Using Swin Transformercited by 1,887
  4. 2021Restormer: Efficient Transformer for High-Resolution Image Restorationcited by 1,627
  5. 2016Beyond a Gaussian Denoiser: Residual Learning of Deep CNN for Image Denoisingcited by 1,358
  6. 2014Image Super-Resolution Using Deep Convolutional Networkscited by 1,224
  7. 2021Uformer: A General U-Shaped Transformer for Image Restorationcited by 866
  8. 2021Multi-Stage Progressive Image Restorationcited by 816
  9. 2021Resolution-robust Large Mask Inpainting with Fourier Convolutionscited by 738
  10. 2018Deep Retinex Decomposition for Low-Light Enhancementcited by 723
  11. 2022Simple Baselines for Image Restorationcited by 708
  12. 2024KAN: Kolmogorov-Arnold Networkscited by 649
38 more
  1. 2019Toward Real-World Single Image Super-Resolution: A New Benchmark and A New Modelcited by 403
  2. 2024MambaIR: A Simple Baseline for Image Restoration with State-Space Modelcited by 352
  3. 2019FFA-Net: Feature Fusion Attention Network for Single Image Dehazingcited by 307
  4. 2024MACE: Mass Concept Erasure in Diffusion Modelscited by 307
  5. 2020Multi-Scale Progressive Fusion Network for Single Image Derainingcited by 260
  6. 2022Vision Transformers for Single Image Dehazingcited by 260
  7. 2019An Underwater Image Enhancement Benchmark Dataset and Beyondcited by 253
  8. 2023Retinexformer: One-stage Retinex-based Transformer for Low-light Image Enhancementcited by 231
  9. 2020Multi-Scale Boosted Dehazing Network with Dense Feature Fusioncited by 230
  10. 2023Learning A Sparse Transformer Network for Effective Image Derainingcited by 205
  11. 2023TF-ICON: Diffusion-Based Training-Free Cross-Domain Image Compositioncited by 182
  12. 2020FcaNet: Frequency Channel Attention Networkscited by 179
  13. 2022Ultra-High-Definition Low-Light Image Enhancement: A Benchmark and Transformer-Based Methodcited by 174
  14. 2022Efficient Frequency Domain-based Transformers for High-Quality Image Deblurringcited by 165
  15. 2023Efficient and Explicit Modelling of Image Hierarchies for Image Restorationcited by 145
  16. 2023Low-Light Image Enhancement with Wavelet-based Diffusion Modelscited by 127
  17. 2024MIGC: Multi-Instance Generation Controller for Text-to-Image Synthesiscited by 124
  18. 2019Frequency Separation for Real-World Super-Resolutioncited by 115
  19. 2021U-shape Transformer for Underwater Image Enhancementcited by 112
  20. 2024Robust Watermarking Using Generative Priors Against Image Editing: From Benchmarking to Advancescited by 110
  21. 2020NH-HAZE: An Image Dehazing Benchmark with Non-Homogeneous Hazy and Haze-Free Imagescited by 108
  22. 2019Dense Haze: A benchmark for image dehazing with dense-haze and haze-free imagescited by 107
  23. 2022LEDNet: Joint Low-light Enhancement and Deblurring in the Darkcited by 96
  24. 2018O-HAZE: a dehazing benchmark with real hazy and haze-free outdoor imagescited by 95
  25. 2024EraseAnything: Enabling Concept Erasure in Rectified Flow Transformerscited by 95
  26. 2023Wavelet-based Fourier Information Interaction with Frequency Diffusion Adjustment for Underwater Image Restorationcited by 88
  27. 2023Embedding Fourier for Ultra-High-Definition Low-Light Image Enhancementcited by 86
  28. 2023Curricular Contrastive Regularization for Physics-aware Single Image Dehazingcited by 83
  29. 2021Intriguing Findings of Frequency Selection for Image Deblurringcited by 81
  30. 2023MB-TaylorFormer: Multi-branch Efficient Transformer Expanded by Taylor Formula for Image Dehazingcited by 81
  31. 2023Pyramid Diffusion Models For Low-light Image Enhancementcited by 67
  32. 2024AddSR: Accelerating Diffusion-based Blind Super-Resolution with Adversarial Diffusion Distillationcited by 66
  33. 2021Structure-Preserving Deraining with Residue Channel Prior Guidancecited by 63
  34. 2025HVI: A New Color Space for Low-light Image Enhancementcited by 63
  35. 2025Set You Straight: Auto-Steering Denoising Trajectories to Sidestep Unwanted Conceptscited by 62
  36. 2024Exploiting Multimodal Spatial-temporal Patterns for Video Object Trackingcited by 57
  37. 2024Kolmogorov-Arnold Transformercited by 56
  38. 2015Seven ways to improve example-based single image super resolutioncited by 51

50 most cited of 89 All 89 (JSON)

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 7.

7 papers.