Zero-Shot Solving of Imaging Inverse Problems via Noise-Refined Likelihood Guided Diffusion Models

TL;DR

This paper introduces a zero-shot diffusion-based framework for imaging inverse problems that does not require retraining for different degradation types, using likelihood-guided noise refinement for improved reconstruction quality.

Contribution

The authors propose a novel likelihood-guided noise refinement method for diffusion models, enabling zero-shot solving of diverse imaging inverse problems without retraining.

Findings

Achieves superior reconstruction quality across multiple inverse problems.

Performs well even at very low sampling rates like 5%.

Enhances inference efficiency with DDIM sampling.

Abstract

Diffusion models have achieved remarkable success in imaging inverse problems owing to their powerful generative capabilities. However, existing approaches typically rely on models trained for specific degradation types, limiting their generalizability to various degradation scenarios. To address this limitation, we propose a zero-shot framework capable of handling various imaging inverse problems without model retraining. We introduce a likelihood-guided noise refinement mechanism that derives a closed-form approximation of the likelihood score, simplifying score estimation and avoiding expensive gradient computations. This estimated score is subsequently utilized to refine the model-predicted noise, thereby better aligning the restoration process with the generative framework of diffusion models. In addition, we integrate the Denoising Diffusion Implicit Models (DDIM) sampling strategy to further improve inference efficiency. The proposed mechanism can be applied to both optimization-based and sampling-based schemes, providing an effective and flexible zero-shot solution for imaging inverse problems. Extensive experiments demonstrate that our method achieves superior performance across multiple inverse problems, particularly in compressive sensing, delivering high-quality reconstructions even at an extremely low sampling rate (5%).