Learning Efficient and Effective Trajectories for Differential Equation-based Image Restoration

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Zhu, Zhiyu, Hou, Jinhui, Liu, Hui, Zeng, Huanqiang, Hou, Junhui
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909641981558784
author Zhu, Zhiyu
Hou, Jinhui
Liu, Hui
Zeng, Huanqiang
Hou, Junhui
author_facet Zhu, Zhiyu
Hou, Jinhui
Liu, Hui
Zeng, Huanqiang
Hou, Junhui
contents The differential equation-based image restoration approach aims to establish learnable trajectories connecting high-quality images to a tractable distribution, e.g., low-quality images or a Gaussian distribution. In this paper, we reformulate the trajectory optimization of this kind of method, focusing on enhancing both reconstruction quality and efficiency. Initially, we navigate effective restoration paths through a reinforcement learning process, gradually steering potential trajectories toward the most precise options. Additionally, to mitigate the considerable computational burden associated with iterative sampling, we propose cost-aware trajectory distillation to streamline complex paths into several manageable steps with adaptable sizes. Moreover, we fine-tune a foundational diffusion model (FLUX) with 12B parameters by using our algorithms, producing a unified framework for handling 7 kinds of image restoration tasks. Extensive experiments showcase the $\textit{significant}$ superiority of the proposed method, achieving a maximum PSNR improvement of 2.1 dB over state-of-the-art methods, while also greatly enhancing visual perceptual quality. Project page: https://zhu-zhiyu.github.io/FLUX-IR/.
format Preprint
id arxiv_https___arxiv_org_abs_2410_04811
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Learning Efficient and Effective Trajectories for Differential Equation-based Image Restoration
Zhu, Zhiyu
Hou, Jinhui
Liu, Hui
Zeng, Huanqiang
Hou, Junhui
Computer Vision and Pattern Recognition
The differential equation-based image restoration approach aims to establish learnable trajectories connecting high-quality images to a tractable distribution, e.g., low-quality images or a Gaussian distribution. In this paper, we reformulate the trajectory optimization of this kind of method, focusing on enhancing both reconstruction quality and efficiency. Initially, we navigate effective restoration paths through a reinforcement learning process, gradually steering potential trajectories toward the most precise options. Additionally, to mitigate the considerable computational burden associated with iterative sampling, we propose cost-aware trajectory distillation to streamline complex paths into several manageable steps with adaptable sizes. Moreover, we fine-tune a foundational diffusion model (FLUX) with 12B parameters by using our algorithms, producing a unified framework for handling 7 kinds of image restoration tasks. Extensive experiments showcase the $\textit{significant}$ superiority of the proposed method, achieving a maximum PSNR improvement of 2.1 dB over state-of-the-art methods, while also greatly enhancing visual perceptual quality. Project page: https://zhu-zhiyu.github.io/FLUX-IR/.
title Learning Efficient and Effective Trajectories for Differential Equation-based Image Restoration
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2410.04811