Quick Bypass Mechanism of Zero-Shot Diffusion-Based Image Restoration

TL;DR

This paper introduces a Quick Bypass Mechanism (QBM) that accelerates zero-shot diffusion-based image restoration by initializing from an intermediate step, combined with a Revised Reverse Process (RRP) to improve stochasticity and consistency.

Contribution

The paper proposes a novel QBM strategy to significantly speed up zero-shot diffusion-based image restoration without sacrificing performance.

Findings

QBM reduces denoising iterations by over 50%.

The combined approach maintains high restoration quality.

Experiments on ImageNet-1K and CelebA-HQ validate effectiveness.

Abstract

Recent advancements in diffusion models have demonstrated remarkable success in various image generation tasks. Building upon these achievements, diffusion models have also been effectively adapted to image restoration tasks, e.g., super-resolution and deblurring, aiming to recover high-quality images from degraded inputs. Although existing zero-shot approaches enable pretrained diffusion models to perform restoration tasks without additional fine-tuning, these methods often suffer from prolonged iteration times in the denoising process. To address this limitation, we propose a Quick Bypass Mechanism (QBM), a strategy that significantly accelerates the denoising process by initializing from an intermediate approximation, effectively bypassing early denoising steps. Furthermore, recognizing that approximation may introduce inconsistencies, we introduce a Revised Reverse Process (RRP), which adjusts the weighting of random noise to enhance the stochasticity and mitigate potential disharmony. We validate proposed methods on ImageNet-1K and CelebA-HQ across multiple image restoration tasks, e.g., super-resolution, deblurring, and compressed sensing. Our experimental results show that the proposed methods can effectively accelerate existing methods while maintaining original performance.