Frequency-Decoupled Diffusion Guidance for Non-Blind Image Deblurring

2026-10-05

AI Summary

This study addresses the inadequate coupled modeling of frequency activation and attenuation in non-blind image deblurring by proposing a frequency-decoupled posterior guidance framework. For the first time, this method explicitly decouples frequency activation from attenuation-aware spectral regularization. It further stabilizes the diffusion sampling process by integrating a KL-regularized path space interpretation with a local trajectory regularizer. To enhance restoration quality, the framework introduces progressive low-to-high frequency scheduling, kernel-derived attenuation maps, and Tweedie approximation. Evaluated on natural image benchmarks, the proposed approach achieves superior PSNR and SSIM performance, demonstrating particularly notable robustness advantages in high-noise scenarios.

Abstract

Pretrained diffusion models provide powerful image priors for training-free posterior sampling in image restoration. To guide this sampling process, frequency-aware methods progressively incorporate measurement information across frequency bands, facilitating coarse-to-fine reconstruction. However, existing methods typically do not explicitly separate frequency activation from degradation-induced attenuation, leaving attenuation differences among inactive frequencies insufficiently modeled. In this work, we propose frequency-decoupled posterior guidance to separate frequency activation from attenuation-aware spectral regularization. Specifically, a progressive low-to-high frequency schedule determines the active measurement band, while a kernel-derived attenuation map defines a selective spectral prior over inactive components. To stabilize the sampling process, we also introduce a local trajectory regularizer that suppresses spatially irregular state-to-clean deviations. For a fixed endpoint energy, we provide a KL-regularized path-space interpretation. In practice, we construct time-dependent guidance through local energy corrections using a Tweedie plug-in approximation. Experiments on natural-image benchmarks demonstrate strong PSNR and SSIM performance across challenging non-blind deblurring settings, even at higher measurement noise levels.

Problem

• Non-blind image deblurring
• Diffusion models
• Frequency-decoupled guidance
• Image restoration
• Posterior sampling

Innovation

• Frequency-Decoupled Guidance
• Non-Blind Image Deblurring
• Diffusion Models
• Spectral Regularization
• Tweedie Approximation