Research map / Image generation / Image restoration
Unveiling Advanced Frequency Disentanglement Paradigm for Low-Light Image Enhancement
Kun Zhou, Xinyu Lin, Wenbo Li, Xiaogang Xu, Yuanhao Cai, Zhonghang Liu, Xiaoguang Han, Jiangbo Lu
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| apply frequency-disentangled coarse-to-fine training (ACCA + LDRM) to existing LLIE models | Restormer · SDSD-out | PSNR (dB): 24.79 -> 32.47 | Improved |
| apply the disentanglement scheme | Retinexformer · SID | PSNR (dB): 24.44 -> 24.64 | Improved |
| apply the disentanglement scheme to all six baselines | MIR-Net, Restormer, LLFlow, SNR, Retinexformer, Diff-L · LOL-v2, SID, SDSD, SMID | PSNR (dB): 20.02 -> 24.19 (MIR-Net, LOL-v2) | Improved |
| use the frequency-disentangled scheme instead of segmentation-guided SKF | SNR · LOL-v2 | PSNR (dB): 21.93 (SKF) -> 24.00 | Improved |
| add low-frequency consistent loss L_i (ablation) | Restormer + ACCA · LOL-v2 | PSNR (dB): 20.21 -> 24.56 | Improved |
| add ACCA and LDRM without L_i (ablation) | Restormer · LOL-v2 | PSNR (dB): 19.94 -> 20.21 | No clear effect |
| freeze ACCA (decoupled optimization) instead of end-to-end training (ablation) | LLFlow · LOL-v2 | PSNR (dB): 27.94 -> 28.90 | Improved |
| replace ACCA with ZeroDCE, Star, PairLIE or IAT as coarse module (ablation) | Restormer · LOL-v2 | PSNR (dB): ZeroDCE 21.53, IAT 23.97 vs ACCA 24.56 | Mixed |
| use four Laplacian pyramid levels rather than 3, 5 or 6 (ablation) | LDRM · LOL-v2 | PSNR (dB) | Improved |
| add the method to existing models | six LLIE baselines · model complexity | parameters: +0.088M params, +2.53 GFLOPS, +0.008s | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- Retinexformer — Retinexformer: One-stage Retinex-based Transformer for Low-light Image Enhancement (ICCV 2023) · better on 3 of 3 shared results
- Restormer — Restormer: Efficient Transformer for High-Resolution Image Restoration (CVPR 2022) · better on 2 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Zero-Shot Low-Light Image Enhancement via Latent Diffusion Models (AAAI 2025 · similarity 0.87)
- Enhancing Low-Light Images: A Synthetic Data Perspective on Practical and Generalizable Solutions (AAAI 2025 · similarity 0.87)
- Interpretable Unsupervised Joint Denoising and Enhancement for Real-World low-light Scenarios (ICLR 2025 · similarity 0.86)
- Multi-Domain Multi-Scale Diffusion Model for Low-Light Image Enhancement (AAAI 2024 · similarity 0.85)
- DPLUT: Unsupervised Low-light Image Enhancement with Lookup Tables and Diffusion Priors (AAAI 2025 · similarity 0.85)
- FRBNet: Revisiting Low-Light Vision through Frequency-Domain Radial Basis Network (NeurIPS 2025 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
