结构表面裂缝数字图像法识别研究综述与前景展望

裂缝反映结构受力状态与安全性、耐久性，是结构现场安全性检测监测以及结构模型试验研究的重要指标之一。现有的人工裂缝识别技术难以满足工程现场与实验室需求，操作中存在测不准、高空多、效率低、记不全等缺点。相比之下，采用数字图像法进行结构表面裂缝识别，具有便捷、自动、定量、准确等优势。文章对结构表面裂缝数字图像法识别研究进行系统综述，对裂缝识别中图像预处理、裂缝识别与提取、裂缝参数计算等重要环节的常见算法进行讨论，阐述采用多视角几何三维重建方法实现裂缝成像修正与拼接、裂缝表达输出的原理与流程，结合实桥案例报道了基于无人机平台的裂缝识别研究与应用，讨论国内首部数字图像法检测规程《工程结构数字图像法检测技术规程》征求意见稿中裂缝检测的相关规定。最后，对结构表面裂缝数字图像法识别研究进行前景展望。

Review and prospect of digital-image-based crack detection of structure surface

Cracks reflecting the stress state, safety and durability of structures are one of the important indicators for on-site safety inspection, monitoring of concrete structures and experimental research of structural model. The existing manual crack detection technology may not meet the needs of engineering projects and laboratories, due to its disadvantages such as inaccurate measurement, too much work at height, low efficiency, and incomplete records. In comparison, the use of digital image method to identify structural surface cracks has the benefits of convenience, automation, quantification and high accuracy. This paper provides a systematic review on the researches of structural surface crack detection by digital image method, and analyzes and discusses those commonly-used algorithms involved in important procedures of crack detection including image preprocessing, crack detection and extraction, and calculation of crack parameters. The principle and process of using multi-view geometric 3D reconstruction methods to realize crack imaging correction and stitching, and crack expression output are discussed. Based on actual bridge cases, the study and application of crack detection using the UAV platform are reported. The relevant regulations for crack detection in the exposure draft of “Technical specification for digital image inspection of engineering structures”, as the first domestic digital image method inspection specification, are introduced. Finally, the prospect of the research on the detection of structural surface defects by digital image method is discussed.