Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2602.11553 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866914570770055168 |
|---|---|
| author | Nguyen, Nam Nguyen, Thinh Bose, Bella |
| author_facet | Nguyen, Nam Nguyen, Thinh Bose, Bella |
| contents | Image denoising aims to remove noise while preserving structural details and perceptual realism, yet distortion-driven methods often produce over-smoothed reconstructions, especially under strong noise and distribution shift. This paper proposes a generative compression framework for perception-based denoising, where restoration is achieved by reconstructing from entropy-coded latent representations that enforce low-complexity structure, while generative decoders recover realistic textures via perceptual measures such as learned perceptual image patch similarity (LPIPS) loss and Wasserstein distance. Two complementary instantiations are introduced: (i) a conditional Wasserstein GAN (WGAN)-based compression denoiser that explicitly controls the rate-distortion-perception (RDP) trade-off, and (ii) a conditional diffusion-based reconstruction strategy that performs iterative denoising guided by compressed latents. We further establish non-asymptotic guarantees for the compression-based maximum-likelihood denoiser under additive Gaussian noise, including bounds on reconstruction error and decoding error probability. Experiments on synthetic and real-noise benchmarks demonstrate consistent perceptual improvements while maintaining competitive distortion performance. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2602_11553 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Perception-based Image Denoising via Generative Compression Nguyen, Nam Nguyen, Thinh Bose, Bella Computer Vision and Pattern Recognition Artificial Intelligence Image denoising aims to remove noise while preserving structural details and perceptual realism, yet distortion-driven methods often produce over-smoothed reconstructions, especially under strong noise and distribution shift. This paper proposes a generative compression framework for perception-based denoising, where restoration is achieved by reconstructing from entropy-coded latent representations that enforce low-complexity structure, while generative decoders recover realistic textures via perceptual measures such as learned perceptual image patch similarity (LPIPS) loss and Wasserstein distance. Two complementary instantiations are introduced: (i) a conditional Wasserstein GAN (WGAN)-based compression denoiser that explicitly controls the rate-distortion-perception (RDP) trade-off, and (ii) a conditional diffusion-based reconstruction strategy that performs iterative denoising guided by compressed latents. We further establish non-asymptotic guarantees for the compression-based maximum-likelihood denoiser under additive Gaussian noise, including bounds on reconstruction error and decoding error probability. Experiments on synthetic and real-noise benchmarks demonstrate consistent perceptual improvements while maintaining competitive distortion performance. |
| title | Perception-based Image Denoising via Generative Compression |
| topic | Computer Vision and Pattern Recognition Artificial Intelligence |
| url | https://arxiv.org/abs/2602.11553 |