Research map / Image generation / Image restoration
DAQE: Enhancing the Quality of Compressed Images by Exploiting the Inherent Characteristic of Defocus
Qunliang Xing, Mai Xu, Xin Deng, Yichen Guo
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 |
|---|---|---|---|
| add CA subnet connecting QENet levels (ablation) | DAQE · DIV2K (BPG QP=37) | PSNR (dB): 32.59 -> 32.69 | Improved |
| add global attention in AGNet (ablation) | DAQE · DIV2K (BPG QP=37) | PSNR (dB): 32.54 -> 32.59 | Improved |
| add local attention in AGNet (ablation) | DAQE · DIV2K (BPG QP=37) | PSNR (dB): 32.51 -> 32.54 | Improved |
| cluster patches by true compression quality (upper bound) instead of estimated defocus (ablation) | DAQE · DIV2K (BPG QP=37) | PSNR (dB): 32.69 -> 32.84 | Improved |
| use defocus-based attention and clustering instead of defocus-blind (ablation) | DAQE · DIV2K (BPG QP=37, exit at level 1) | PSNR (dB): 32.47 -> 32.56 | Improved |
| use defocus-based clustering instead of defocus-blind (ablation) | DAQE · DIV2K (BPG QP=37, exit at level 1) | FPS: 1.85 -> 2.81 | Improved |
| cluster by defocus instead of wavelet-energy frequency (ablation) | DAQE · DIV2K (BPG QP=37) | PSNR (dB): 32.52 -> 32.69 | Improved |
| feed clean wavelet energy of raw patches to DAQE-Freq (ablation) | DAQE-Freq · DIV2K (BPG QP=37) | PSNR (dB): 32.52 -> 32.62 | Improved |
| use defocus-based dynamic exit instead of fixed or random exit levels (ablation) | DAQE · DIV2K (BPG QP=37) | PSNR-FPS tradeoff | Improved |
| train DENet on compressed images instead of retraining DMENet with compression | defocus estimator · CUHK (BPG QP=37) | accuracy (%): 76.30 -> 81.73 | Improved |
| compression artifacts degrade defocus estimator (DMENet) | DMENet · CUHK (BPG QP=37) | accuracy (%): 76.30 -> 70.96 | Hurt |
| use DAQE instead of CBDNet | image enhancer · DIV2K (BPG QP=37) | PSNR (dB): 32.40 -> 32.69 | Improved |
Closest papers by meaning
- Enhancing Quality of Compressed Images by Mitigating Enhancement Bias Towards Compression Domain (CVPR 2024 · similarity 0.87)
- Quad-Pixel Image Defocus Deblurring: A New Benchmark and Model (CVPR 2025 · similarity 0.86)
- Breaking the Multi-Enhancement Bottleneck: Domain-Consistent Quality Enhancement for Compressed Images (TPAMI 2025 · similarity 0.86)
- Residual Diffusion Deblurring Model for Single Image Defocus Deblurring (AAAI 2025 · similarity 0.85)
- Prior and Prediction Inverse Kernel Transformer for Single Image Defocus Deblurring (AAAI 2024 · similarity 0.84)
- Multi-Focus Image Fusion via Explicit Defocus Blur Modelling (AAAI 2025 · similarity 0.84)
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.
