SCoRe: Clean Image Generation from Diffusion Models Trained on Noisy Images
Yuta Matsuzaki, Seiichi Uchida, Shumpei Takezaki
TLDR
SCoRe is a training-free method that uses spectral regeneration to produce clean images from diffusion models trained on noisy datasets, improving quality.
Key contributions
- Suppresses corrupted high-frequency components in generated images.
- Regenerates high-frequency details using SDEdit, leveraging spectral bias.
- Derives a theoretical mapping for cutoff frequency and SDEdit timestep via RAPSD.
- Achieves cleaner images from noisy-trained diffusion models without retraining.
Why it matters
Diffusion models trained on noisy data often produce artifacts, degrading image quality. SCoRe offers a novel, training-free approach to generate clean images from such models. This significantly improves output quality without costly retraining or fine-tuning.
Original Abstract
Diffusion models trained on noisy datasets often reproduce high-frequency training artifacts, significantly degrading generation quality. To address this, we propose SCoRe (Spectral Cutoff Regeneration), a training-free, generation-time spectral regeneration method for clean image generation from diffusion models trained on noisy images. Leveraging the spectral bias of diffusion models, which infer high-frequency details from low-frequency cues, SCoRe suppresses corrupted high-frequency components of a generated image via a frequency cutoff and regenerates them via SDEdit. Crucially, we derive a theoretical mapping between the cutoff frequency and the SDEdit initialization timestep based on Radially Averaged Power Spectral Density (RAPSD), which prevents excessive noise injection during regeneration. Experiments on synthetic (CIFAR-10) and real-world (SIDD) noisy datasets demonstrate that SCoRe substantially outperforms post-processing and noise-robust baselines, restoring samples closer to clean image distributions without any retraining or fine-tuning.
📬 Weekly AI Paper Digest
Get the top 10 AI/ML arXiv papers from the week — summarized, scored, and delivered to your inbox every Monday.