测绘科学技术学报2025，Vol.41Issue(6)：583-588,6.DOI:10.3969/j.issn.1673-6338.2025.06.005
水工建筑物贴近摄影测量表面缺陷检测方法
Surface Defect Detection Method for Hydraulic Structures Based on Close-range Photogrammetry
摘要
Abstract
The surface structures of hydraulic engineering structures are inherently complex and irregular,making them vulnerable to natural factors that lead to defects such as cracks and exposed reinforcements.To achieve pre-cise defect detection results,a method for extracting surface features and identifying defects by integrating UAV-based close-range photogrammetry with a deep neural network model is proposed.Firstly,multi-view images of the hydraulic structures' surfaces are acquired by using a UAV equipped with a photogrammetric camera,and an im-age fusion dataset is constructed.Then,an improved U-Net-based fully connected neural network is employed to process the image data.Finally,the attention mechanism is used to accurately identify and annotate the defect fea-tures of the hydraulic structures.Taking the spillway of a certain hydropower station as the experimental object,the test results demonstrate that the proposed approach can effectively identify the defect features such as spalling,cracks,and exposed reinforcements of the spillway.This method can achieve refined detection of surface defects,providing a reliable foundation for maintenance,reinforcement,and disaster prevention.关键词
贴近摄影测量/U-Net模型/水工建筑物/表面缺陷检测/神经网络Key words
close-range photogrammetry/U-Net/hydraulic structures/surface defect detection/neural network分类
天文与地球科学引用本文复制引用
苏秀永,沈默..水工建筑物贴近摄影测量表面缺陷检测方法[J].测绘科学技术学报,2025,41(6):583-588,6.基金项目
浙江华东测绘与工程安全技术有限公司科研项目(ZKY2023-CA-02-02). （ZKY2023-CA-02-02）
