东南大学学报（自然科学版）2026，Vol.56Issue(10)：1373-1383,11.DOI:10.3969/j.issn.1001-0505.2026.10.001
基于机器视觉法的结构表观损伤检测研究综述
A review of research on structural surface damage inspection based on computer vision methods
摘要
Abstract
Timely damage inspection is critical to ensuring structural safety.In recent years,computer vision,with its objectivity and efficiency,has become a core alternative to manual visual inspection.To systemati-cally review research progress and trends in this field,this paper presents an overview of computer vision-based structural surface damage inspection.It first summarizes common image acquisition devices,then introduces various damage recognition techniques,and elaborates on data bottlenecks in model training and corresponding solutions.It further discusses damage localization and quantification via 3D reconstruction,as well as damage condition assessment strategies integrating specifications,damage mechanisms,and data-driven methods.Finally,cutting-edge advances such as multimodal large language models are reviewed.Focusing on both engineering applications and the technical details and evolution of related algorithms,this pa-per provides a reference for intelligent structural inspection and assessment.关键词
机器视觉/损伤检测/损伤评估/深度学习/三维重建/大模型Key words
computer vision/damage inspection/damage assessment/deep learning/3D reconstruction/large models分类
建筑与水利引用本文复制引用
王硕,聂鑫,樊立龙,肖飞知..基于机器视觉法的结构表观损伤检测研究综述[J].东南大学学报（自然科学版）,2026,56(10):1373-1383,11.基金项目
国家自然科学基金资助项目(52325802). （52325802）
