Research Progress on All-in-one Image Restoration Models under Multiple Harsh Imaging Environments
JIAO Sihang
GAO Lin
GAO Shijia
LIU Huaguo
Abstract：To enhance the generalization and adaptability of image restoration models under multiple harsh imaging environments,the research progress on all-in-one image restoration models was discussed systematically.Firstly,the current situation of six categories of all-in-one image restoration models under multiple harsh imaging environments was analyzed.Secondly,the characteristics of these models were summarized.Thirdly,the limitations of these models were discussed.Finally,the development trends were envisaged.It was found that current all-in-one image restoration models had improved image clarity and multi-scene applicability.However,they still suffered from limitations such as poor performance on real-world degraded images,low interpretability,and high model complexity.Future improvements could be achieved through real degradation modeling and self-supervised learning to enhance restoration performance;integrating attention mechanisms,physical modeling,and physics-informed networks to improve interpretability;and adopting modular design and model compression to enable lightweight deployment.This study provided important reference for the research and application of high-performance,general-purpose image restoration models.
Keywords：harsh imaging environmentdegraded imagesdeep learningall-in-on restorationcomputer vision
Publication Date：2025-12-30
Online Publishing Date：2025-12-10(First online date of this platform, not the publication date of the document)
Pages：6( 552-557 )